Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Genotype-specific phenotypic behaviour of auxotrophic and prototrophic yeast gene deletion collections Acton, Erica 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_february_acton_erica.pdf [ 2.77MB ]
Metadata
JSON: 24-1.0340607.json
JSON-LD: 24-1.0340607-ld.json
RDF/XML (Pretty): 24-1.0340607-rdf.xml
RDF/JSON: 24-1.0340607-rdf.json
Turtle: 24-1.0340607-turtle.txt
N-Triples: 24-1.0340607-rdf-ntriples.txt
Original Record: 24-1.0340607-source.json
Full Text
24-1.0340607-fulltext.txt
Citation
24-1.0340607.ris

Full Text

GENOTYPE-SPECIFIC PHENOTYPIC BEHAVIOUR OF AUXOTROPHIC AND PROTOTROPHIC YEAST GENE DELETION COLLECTIONS by  Erica Acton  B.Sc., Queen’s University, 2011  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Genome Science and Technology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  December 2016  © Erica Acton, 2016 ii  Abstract The Yeast Knockout (YKO) collection has provided functional annotations from thousands of genome-wide screens. As an unintended consequence however, ~90% of gene annotations are derived from a single genotype. The nutritional auxotrophies in the YKO are of particular concern as they have phenotypic consequences. To address this issue, repaired ‘prototrophic’ versions of the YKO collection have been constructed; the first by introducing an ARS-CEN plasmid carrying wildtype copies of the auxotrophic markers (Plasmid-Borne, PBprot), and the second by backcrossing (Backcrossed, BCprot) to a strain wildtype for the auxotrophies. To systematically assess the impact of the auxotrophies, genome-wide fitness profiles of the prototrophic and auxotrophic YKO collections were compared across a diverse set of drug and environmental conditions. Comparative fitness profiling for the prototrophic collections revealed genotypic and strain-construction-specific phenotypes. The PBprot collection exhibited fitness defects associated with plasmid maintenance, while the BCprot collection’s fitness profiles were compromised due to strain loss resulting from nutrient selection steps during strain construction. The repaired prototrophic versions of the YKO collection did not restore wildtype behaviour and had additional experimental liabilities. Neither prototrophic collection compensated for gaps in gene annotation resulting from the auxotrophic YKO genetic background. To remove marker bias and expand the experimental scope of current deletion libraries, construction of a bona fide prototrophic collection from a wildtype strain will be required.  iii  Preface A version of chapter 2 has been submitted for publication: Acton E, Lee A, Zhao PJ, Flibotte S, Sinha S, Chiang J, Flaherty P, Nislow C, Giaever G. “Comparative functional genomic screens of three yeast deletion collections reveal unexpected effects of genotype in response to diverse stress”. Portions of the introductory text including Table 1 are also from this manuscript. I, initially with Dr. Lee, carried out all biological experiments, participated in the design of the study, participated in the microarray analysis, and drafted the manuscript. Dr. Flibotte, Dr. Giaever, and I conceived the Bar-seq and microarray analysis. Dr. Flaherty provided statistical support. Dr. Sinha conceived of the Bar-seq methodology, performed the sequencing and originally drafted 2.4.1.5. Dr. Flibotte performed the Bar-seq alignment, generated the count and mismatched tag data, and originally drafted 2.4.2.4. J Chiang participated in the design of the microarray experiments. Dr. Nislow and Dr. Giaever were the supervisory authors on this project and conceived of the study, participated in its design, and helped to draft the manuscript. All authors read, edited and approved the final manuscript. Thank you to Kelsey Leduc for the graphic design of Figure 13.  iv  Table of Contents Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iii Table of Contents ......................................................................................................................... iv List of Tables ............................................................................................................................... vii List of Figures ............................................................................................................................. viii List of Abbreviations .....................................................................................................................x Acknowledgements ..................................................................................................................... xii Dedication ................................................................................................................................... xiii Chapter 1: Introduction ................................................................................................................1 1.1 The Yeast Knockout (YKO) collections as an important tool for functional annotation 1 1.2 Evidence that auxotrophic markers are not benign ......................................................... 3 1.3 Prototrophic yeast deletion collections ........................................................................... 6 1.3.1 Restoration of prototrophy by plasmid complementation .......................................... 6 1.3.2 Restoration of prototrophy by backcrossing to a wildtype strain ............................... 7 1.4 Competitive fitness profiling: a tool to compare auxotrophic and prototrophic deletion collection ..................................................................................................................................... 9 1.4.1 Background ................................................................................................................. 9 1.4.2 Methods to assess competitive fitness profiling ....................................................... 10 1.4.2.1 Microarray......................................................................................................... 10 1.4.2.2 Barcode sequencing (Bar-seq) .......................................................................... 11 1.5 Thesis rationale ............................................................................................................. 11 Chapter 2: Exploring the impact of prototrophic restoration on genome-wide phenotypes 14 v  2.1 Introduction ................................................................................................................... 14 2.2 Results ........................................................................................................................... 14 2.2.1 Genetic roster of each deletion collection ................................................................. 14 2.2.2 Comparative fitness profiling ................................................................................... 19 2.2.2.1 Nutrient limiting conditions .............................................................................. 21 2.2.2.1.1 Genotype-specific phenotypic behaviour of the cpa1∆ strain .................... 33 2.2.2.2 Drug and small molecule stress conditions ....................................................... 37 2.2.3 Bar-seq validation of microarray data ....................................................................... 40 2.3 Discussion ..................................................................................................................... 45 2.4 Materials and methods .................................................................................................. 47 2.4.1 Experimental methods .............................................................................................. 47 2.4.1.1 Yeast deletion strains and media preparation ................................................... 47 2.4.1.2 Construction of gene deletion collection pools ................................................. 47 2.4.1.3 Competitive fitness profiling ............................................................................ 48 2.4.1.3.1 Synthetic media HOP screens ..................................................................... 48 2.4.1.3.2 Chemical HOP screens ................................................................................ 48 2.4.1.3.3 Genomic DNA extraction and hybridization to an Affymetrix TAG4 microarray 49 2.4.1.4 cpa1∆ strain phenotype validation .................................................................... 49 2.4.1.5 Library preparation ........................................................................................... 49 2.4.2 Data analysis ............................................................................................................. 50 2.4.2.1 Array normalization and pre-processing ........................................................... 50 2.4.2.2 Fitness defect scores ......................................................................................... 51 vi  2.4.2.3 Gene Ontology (GO) enrichment analysis ........................................................ 52 2.4.2.4 Bar-seq analysis ................................................................................................ 52 2.4.2.5 Similarity of ordered gene sets ......................................................................... 53 Chapter 3: Conclusion .................................................................................................................54 3.1 Summary ....................................................................................................................... 54 3.2 Future directions ........................................................................................................... 56 References .....................................................................................................................................60 Appendices ....................................................................................................................................71 Appendix A GO enrichments and gene lists from chapter 2 .................................................... 71 Appendix B Access to data ....................................................................................................... 80  vii  List of Tables Table 1 Genotypes of gene deletion collections used in this study ................................................ 8 Table 2 Drug and media condition replicates assayed per deletion collection. ............................ 20 Table 3 Significantly enriched GO biological processes for 882 gene deletion strains missing from the BCprot collection (Bonferroni-corrected P-values) ......................................................... 71 Table 4 Gene deletion strains present in the HLU signature ........................................................ 74 Table 5 Significantly enriched GO biological processes for 73 gene deletion strains in the HLU signature (Bonferroni-corrected P-values) ................................................................................... 79  viii  List of Figures Figure 1 ECDFs for mean batch-corrected control arrays ............................................................ 15 Figure 2 Deletion strain overlap of the 3 deletion collections ...................................................... 16 Figure 3 Relative proportion of slow growing strains present and missing from each collection 17 Figure 4 GO enrichment map of the 882 gene deletion strains absent in the BCprot collection .... 18 Figure 5 Relative strain presence in YPD is consistent with that observed in SC media. ............ 19 Figure 6 Hierarchical clustering of FD scores for YKOaux and PBprot collections in synthetic dropout media ............................................................................................................................... 22 Figure 7 PBprot fitness signature revealed in conditions requiring expression of the pHLUM plasmid .......................................................................................................................................... 23 Figure 8 The prototrophic collections in HIS- dropout media (+/–HLU signature) ..................... 25 Figure 9 The prototrophic collections in URA- dropout media (+/–HLU signature) ................... 26 Figure 10 The prototrophic collections in LEU- dropout media (+/–HLU signature) ................. 27 Figure 11 Microarray fitness profiles for 4 synthetic dropout conditions .................................... 29 Figure 12 PBprot and YKOaux, but not BCprot deletion strains exhibit fitness defects under mitochondrial stress ...................................................................................................................... 31 Figure 13 Collection-specific phenotypes of the cpa1∆ strain across synthetic media conditions....................................................................................................................................................... 33 Figure 14 Competitive fitness profiling of 3 deletion collections in synthetic media lacking arginine ......................................................................................................................................... 34 Figure 15 Crosstalk between the arginine and pyrimidine metabolic pathways via carbamoyl phosphate ...................................................................................................................................... 35 ix  Figure 16 Rescue of the cpa1∆ strain with arginine supplementation is dependent on genetic background .................................................................................................................................... 36 Figure 17 Competitive fitness profiling of 3 deletion collections in cisplatin .............................. 37 Figure 18 Competitive fitness profiling of 3 deletion collections in FCCP ................................. 39 Figure 19 Competitive fitness profiling of 3 deletion collections in a cationic quinolone ........... 40 Figure 20 Strain presence assessed by microarray is consistent, but less sensitive compared with Bar-seq .......................................................................................................................................... 42 Figure 21 Microarray fitness profiles were independently confirmed by Bar-seq ....................... 44  x  List of Abbreviations ∆ - Deletion ADE- - Synthetic Dropout Media lacking adenine ARG- - Synthetic Dropout Media lacking arginine ARS – Autonomously Replicating Sequence Bar-seq – Barcode Sequencing BCprot – Backcrossed (prototrophic) CCCP - Carbonylcyanide-3-chlorophenylhydrazone CEN – Yeast Centromere DDR – DNA-damage response ECDF – Empirical Cumulative Distribution Function FCCP - Carbonylcyanide-p-trifluoromethoxyphenylhydrazone FD – Fitness Defect GAAC – General Amino Acid Control GO – Gene Ontology HIS- - Synthetic Dropout Media lacking histidine HLU – Histidine, Leucine, and Uracil KanMx – Geneticin resistance selection marker LEU- - Synthetic Dropout Media lacking leucine LYS- - Synthetic Dropout Media lacking lysine MET- - Synthetic Dropout Media lacking methionine MM – Minimal Media ORF – Open Reading Frame xi  PBprot – Plasmid-Borne (prototrophic) pHLUM – Plasmid containing wildtype copies of HIS3, LEU2, URA3, MET15 SC – Synthetic Complete Media SER- - Synthetic Dropout Media lacking serine T0 – Time Zero (no growth) THR- - Synthetic Dropout Media lacking threonine TRP- - Synthetic Dropout Media lacking tryptophan URA- - Synthetic Dropout Media lacking uracil YKOaux – Yeast Knockout (auxotrophic) YPD – Rich Media (Dextrose as carbon source) YPG – Rich Media (Glycerol as carbon source) YPGal – Rich Media (Galactose as carbon source) xii  Acknowledgements I am grateful to my supervisor, Corey Nislow, for his support and confidence in me, as well as the opportunity to advance my scientific knowledge through discussion, timely advice, and the chance to take additional courses. I am thankful to Guri Giaever for her encouragement and enthusiasm about our science and the opportunity to learn new skills. I am grateful to my committee members, Chris Loewen and Sara Mostafavi, for their support and guidance. Thank you to the NSERC CREATE grant for a GSAT Research Rotation Award, and the staff and students of the hosting GSAT rotation laboratories. Special thanks to Martin Hirst and Michelle Moska, Carl Hansen, and Corey Nislow for the insights afforded in exploring new technologies and projects. I thank all present and past members of the Nislow-Giaever laboratory: Amy Lee, Jennifer Chiang, Elisa Wong, Sunita Sinha, Grant Tran, Seth Tigchelaar, Sean Formby, Stephane Flibotte, Mauricio Neira, Mark Mathew. Special thanks to Amy for her mentorship, friendship, cheerleading, and for always having my back. Thanks to Sunita, Jen, and Elisa for their advice and support over numerous lunches, tea breaks and beers. Thanks to Grant for the infinite amounts of moral support and coffee that goes into being a great officemate. I am also grateful to my friend-colleague Analise Hofmann for her support and company throughout the scientific journey.  I am indebted to Robert Lee for his unconditional support for my academic and career pursuits. Finally, thank you to my family and friends for their support in the form of cards, care packages, phone dates, and visits during this long-distance endeavor.   xiii  Dedication  I dedicate this thesis to my loving family; distant, but not far in my heart.  1  Chapter 1: Introduction 1.1 The Yeast Knockout (YKO) collections as an important tool for functional annotation Saccharomyces cerevisiae has a plethora of genetic traits that make it a powerful tool to determine gene function and its homology with other organisms allows extrapolation of these functions to higher eukaryotes. Baker’s yeast short lifespan, ability to stably grow in a haploid or diploid state, and its propensity for homologous recombination along with the ability to integrate DNA chromosomally or carry a plasmid made it a maven for molecular biology. In laboratory strains, the deletion of amino acid and nucleotide biosynthetic genes such as URA3 and LEU2 or the addition of genetic cassettes that confer antibiotic resistance such as KanMx made for facile design and selection of genetic manipulations in a relatively inexpensive manner by media selection. With the availability of Saccharomyces cerevisiae’s genome sequence came a list of putative genes homologous to human disease genes and the potential of determining a function for every gene [1, 2].  The combination of powerful genomic tools and a well-annotated genome sequence made experimental manipulations via reverse genetics straightforward; enabling PCR-mediated gene disruption for the precise start-to-stop deletion of a gene based on sequence microhomology [3]. This technology facilitated the construction of the Yeast Knockout (YKOaux) collection, a genome-wide set of strains each carrying a precise deletion for nearly all 6,000 open reading frames (ORFs) in the yeast genome [4, 5]. The yeast deletion collection is the first complete and systematically constructed deletion collection available for any organism and is a valuable resource for functional genomic studies [4, 5].   2  The yeast knockout collection is composed of ~21,000 deletion strains constructed in four strain backgrounds, derived from the sequenced parental strain S288c: (1) BY4741 MATa his3∆1 leu2∆0 met15∆0 ura3∆0, (2) BY4742 MATα his3∆1 leu2∆0 lys2∆0 ura3∆0 haploid strains, (3) BY4743 homozygous deletion strains carrying the deletions in ~5,000 nonessential genes, and (4) BY4743 heterozygous strains, each strain deleted for a single copy of the ~1,000 essential and ~5,000 nonessential genes [4-6]. Each open reading frame was precisely deleted from start-to-stop and replaced by homologous recombination with a KanMx dominant drug resistance cassette containing two 20bp molecular barcodes that uniquely identify each strain. The barcodes are flanked by common primer sequences allowing for PCR amplification of the barcodes. The abundance of molecular barcodes is assessed by microarray hybridization [7, 8] or sequencing [9-11], resulting in a quantitative metric of relative strain fitness. In this manner, all genes required for growth in a condition of choice can be identified in a single assay. The barcodes also provide concrete evidence of strain presence safeguarding against plate mapping and tracking errors that can arise in the physical handling of the ~60 96-well plate format of the collections. The YKOaux collection has greatly expanded our understanding of gene function and the cellular response to perturbation through comprehensive screens performed in thousands of different environmental and drug conditions [4, 5]. Thus far, over 12,000 YKO genome-wide studies have been published, surveying the cellular response to a vast array of environmental and drug conditions, and providing detailed characterization of gene-environment relationships [5, 6, 12-15]. In addition, the YKOaux collection inspired the construction of ‘designer’ genomic collections and the development of additional techniques such as synthetic genetic array (SGA) [16], green fluorescent protein (GFP) [17], yeast tandom affinity purification (TAP-tagged) [18], and epistatic miniarray profile (EMAPs) libraries [19]. As a result, in the 15 years since the 3  completion of YKOaux collection, the proportion of the genome with functional annotation has increased from approximately 30% to 90% [1, 2, 6, 20, 21].  In the age of functional genomics, during which the YKOaux collections were generated, it was necessary to standardize a strain background that could be interrogated in a high-throughput manner as well as facilitate bulk genetic manipulations. These haploid and diploid collections contain auxotrophic markers (ie. deletion in genes required for amino acid and nucleotide biosynthesis) to allow for selection and counter-selection of transformants (in the case of URA3). Strains that are auxotrophic are missing a key enzyme in a biological pathway and are unable to synthesize a respective amino acid or nucleotide. At the time the YKOaux collections were made, supplementation of these amino acids or nucleotides to the media to allow growth was considered to be physiologically comparable to a wildtype phenotype [6].   1.2 Evidence that auxotrophic markers are not benign An auxotrophic derivative of S288c, an S. cerevisiae wildtype strain originally isolated by Robert Mortimer [22], was chosen as the parent strain of the YKOaux collection to facilitate functional genomics. The auxotrophic markers in the YKO (his3∆1, leu2∆0, met17∆0, ura3∆0, lys2∆0) were considered required features for genetic studies at the time of construction and have long been considered inert. A decade of functional genomic, proteomic and metabolomic studies has however firmly established the impacts of the YKO auxotrophies on cellular physiology. Amino acids are central players in nearly all biosynthetic pathways and maintaining appropriate balance is critical to cell growth, and for coordinating the multiple layers of metabolic control. Simple media supplementation not only does not fully compensate for the auxotrophies, but such complementation can have unexpected consequences [10, 23-25].  4   Altering media composition is not a solution to rescuing auxotrophic phenotypes. Even in rich media (YPD), the growth rate of the BY4741 strain, a YKOaux genotype, is 10-15% less than the parental strain and cultures exhibit decreased biomass [23]. Starvation for auxotrophic requirements has been shown to be fundamentally different than ‘natural starvation’ for nitrogen, carbon, phosphorus and sulfur which might be expected in the wild; the cell does not undergo cell cycle arrest, a self-preserving mechanism that yeast has evolved to survive nutrient scarcity [10]. Instead, the cell continues to grow despite being starved for the auxotrophic nutrient, resulting in cell death. To exacerbate the situation, transporters required to take up the supplementing nutrient may not be expressed in nitrogen-rich media [24]. The discrepancies produced can be significant: starvation of the MATα collection showed that the half-life for auxotrophic yeast is 337 or 28 hours for when starved for phosphorus or leucine, respectively [10]. This ten-fold difference in survival may have profound, unanticipated biological consequences. For example, chronological lifespan studies can be significantly affected depending on which media nutrient is exhausted first, and whether the limiting nutrient in the media was ‘natural’ or ‘unnatural’ [25-27].  Differing auxotrophic and prototrophic phenotypes have been attributed to almost all commonly used amino acid and nucleotide-related biosynthetic markers, individually as when combined in different genetic backgrounds [10, 24, 28-30]. Using proper controls is paramount for interpreting mutant phenotypes. As an example of the impact of a single auxotrophic marker, an increased growth phenotype attributed to an ath1∆ URA3 strain is actually due to a uracil auxotrophy of the ATH1 ura3∆ control strain [30]. Another study gives an example of the confusion resulting from the four standard auxotrophic markers from the MATa genetic 5  background, a screen for acetic acid resistant strains resulted in 23 hits, 11 of which were retested but could not be recapitulated in a prototrophic background [31]. In the pathogenic yeast Candida albicans, the widely used uracil auxotrophy renders these strains non-pathogenic [32]. Environment also affects media supplementation - weak acid stress inhibits external aromatic amino acid uptake; auxotrophic strains, which require additional supplementation, were found to be more sensitive than prototrophic strains to high levels of acetic acid [28]. Auxotrophic mutants ‘resistant’ to acetic acid had deletions whose downstream effects allowed for the uptake of nutrients, bringing these deletion strains to the sensitivity level of a prototrophic strain rather than a truly resistant phenotype [31].  It is therefore important to consider these potential synthetic effects of auxotrophies and to appreciate that in the YKOaux collection each deletion background is a quadruple mutant, and that there are complex epistatic interactions with these deletions [29, 33]. Additionally, any single-gene deletions present in the YKOaux strains may be sufficient to induce additional selection pressure to evolve secondary site mutations that may reflect genome evolution rather than random mutation [34]. Auxotrophies can have profound effects on metabolic regulation, altered metabolite pools, and gene expression which can complicate data interpretation [33, 35]. In the worst case scenario, unappreciated auxotrophic effects can lead to incorrect assignment of gene function, wherein resistant or sensitive phenotypes cannot be recapitulated in the prototrophic background [28, 30].  Typically, these studies have been anecdotal on the scale of individual strains, or a handful of strains that relate to a specific pathway or cellular function, and not at a genome-wide scale. Taken together, these observations confirm the need to systematically examine the impact of unanticipated auxotrophic bias in YKOaux studies. 6  1.3 Prototrophic yeast deletion collections Two prototrophic versions of the YKOaux MATa collection have been constructed, in part to address potential confounding effects of auxotrophy and importantly, allow the interrogation of basic environments that were previously restricted. Indeed, while chemical genomics is providing novel chemical environments that perturb gene function, information is still missing about natural environments yeast encounter outside the laboratory and how different strains would respond to such ‘natural’ stresses. Since annotating gene function is partially dependent on observing a phenotype that deviates from wildtype, this additional assayable space will help to annotate the function of the approximately 700 as yet uncharacterized open reading frames [21]. Prototrophic strains also enable metabolomic studies without exogenous nutrients in the media, simplifying analysis. Prior metabolomics with the MATa YKOaux collection was likely confounded by amino acids in SC media [36].  1.3.1 Restoration of prototrophy by plasmid complementation In the first case, the ‘Plasmid-Borne’ (PBprot) prototrophic collection, auxotrophies were genetically complemented in the haploid MATa collection by introducing a single-copy ARS-CEN plasmid carrying wildtype copies of the HIS3, LEU2, URA3 and MET15 genes (pHLUM) [29]. The initial study introducing this collection further validated the need for a prototrophic collection by comparing the growth rate of a prototrophic strain with that of strains deleted for combinatorial permutations of the auxotrophic markers in synthetic minimal media solely supplemented with the nutrient that complemented their respective auxotrophies. This comparison allowed the quantification of the most deleterious auxotrophic marker in the context of growth defects (leu2∆0) and highlighted that the unpredictable variance in growth rates 7  implied complex interactions between the selection markers [29]. This effect was reduced, but not eliminated in SC by the addition of amino acids. The authors’ also demonstrated that the deletion strains had unequal consumption of nutrients in a SC batch culture, which limited the growth of these strains, in order: the ∆met15 strain, followed by ∆leu2, ∆his3, then ∆ura3 [29]. However, the most dramatic finding resulted from the addition of the pHLUM plasmid to 370 strains in the Tet-promoters Hughes collection (yTHC), a set of essential strains whose promoter can be regulated (in principal) based on the concentration of doxycycline in media. Restoration of prototrophy rescued the lethal phenotype of 13 strains [29]; the auxotrophic background had been responsible for the annotation of these genes as being essential. Only two of the deletions rescued had been reported to be synthetically lethal with an auxotrophic marker [37]. While the utility of prototrophy in augmenting unwanted phenotypic variation resulting from an auxotrophic background was clearly demonstrated, no genome-wide demonstration of the PBprot collection has been performed. 1.3.2 Restoration of prototrophy by backcrossing to a wildtype strain In the second ‘Backcrossed’ (BCprot) collection, the auxotrophic markers were repaired using an SGA-based methodology [16], by backcrossing the MATa collection to a strain prototrophic for the auxotrophic markers and carrying deletions in the arginine (CAN1) and lysine (LYP1) transporters (ACY742:MATα can1Δ::STE2pr-SpHIS5 his3Δ1 lyp1Δ0) [38, 39]. Following SGA and haploid conversion, minimal media was then used for selection of prototrophic haploid deletion strains (Table 1). The genes replacing the auxotrophies were returned to their native chromosomal locations with the exception of STE2pr-SpHIS5 (S. pombe gene used to prevent gene conversion of the HIS3 locus). This collection allowed evaluation of fitness in 28 different carbon and nitrogen sources that would otherwise have been unassayable in the original YKO 8  auxotrophic background [38]. The study showed interactions between nutrients that would have been masked in supplemented media, because amino acids can be used as carbon and nitrogen sources. The study boasted finding an additional 385 galactose-sensitive mutants relative to other studies [5, 38, 40, 41], which were presumably related to prototrophy as well as the seven additional nitrogen/galactose conditions assayed. This large cohort of genes has not been independently confirmed. Additionally, a unique fitness profile was determined for the ∆fmp32 strain that suggested it was involved in the respiratory response when proline was being used as a nitrogen source [38]. This study therefore showed the benefit of prototrophy for the annotation of uncharacterized strains in new conditions. The BCprot collection was only compared to other collections in the context of hits for galactose-sensitive mutants.  Table 1 Genotypes of gene deletion collections used in this study Strain Background Genotype Source YKOaux BY4743 MATa/a his3Δ1/his3Δ1 leu2Δ0/leu2Δ0 LYS2/lys2Δ0 met15Δ0/MET15 ura3Δ0 /ura3Δ0 Giaever, et al., 2002; Winzeler, et al., 1999. BCprot BY4741 MATa can1Δ::STE2pr-SpHIS5 his3Δ1 lyp1Δ0 VanderSluis, et al., 2014., Gibney, et al., 2013. PBprot BY4741 MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 + pHLUM (Addgene ID 40276) Muellder, et al., 2012.   9  1.4 Competitive fitness profiling: a tool to compare auxotrophic and prototrophic deletion collection 1.4.1 Background Competitive fitness profiling is a tool that allows changes in relative strain abundance to be measured genome-wide, simultaneously, between control and treatment conditions. Briefly, strain identification is facilitated by the two molecular barcodes (uptag and downtag) that uniquely identify each deletion mutant. These tags are extracted after competitive growth in a condition of interest, PCR-amplified and hybridized to a microarray chip or sequenced. Strains sensitive or resistant in the treatment condition relative to the control can be identified for further follow-up, or holistic cellular responses can be determined for the entire set of strains in that condition (2.4.1.3). Fitness profiling determines the direct effect of a gene deletion on the cell’s fitness, which makes it an advantageous tool to determine gene function. Fitness profiling can also be used to evaluate a drug with an unknown mechanism of action by observing which cellular processes are perturbed. For example, if strains deleted for genes involved in DNA damage repair were sensitive to a compound, we might consider the compound to be a DNA-damaging agent. Conversely, if we know that a drug is a DNA-damaging agent, but we have a fitness defect for a deletion strain whose gene function is unknown, we might consider that gene is necessary to buffer the effects of DNA damage. Indeed, drugs or environments with similar profiles tend to share related mechanisms or disturb similar biological processes, and thus ‘guilt-by-association’ can be used to infer gene function or compound mechanisms [12-14, 42, 43].   10  Fitness profiling has a few assumptions. The assay relies on an equal starting representation for each strain. Thus, before beginning these assays a pool of deletion strains had to be carefully constructed separately for each deletion collection. Any strains that have significant fitness defects in the absence of perturbation will be challenging to assay. In some cases, slow-growing strains can be spiked in separately in the pool to reach the same approximate starting optical density (OD) as the average strain. Similarly after a pool is created, each sample must have a sufficient number of starting cells to avoid sampling error, since that error is propagated through outgrowth it can result in erroneous, disproportionate strain representation [8]. Additionally, the number of inoculations (ie. sample transfers) the pool undergoes should be minimized, as at this step sampling error due to liquid transfers can occur. Nonetheless, a sufficient number of generations must be used in order to detect strain sensitivity. Consequently, sampling steps and the number of generations must be balanced in the experimental design. Furthermore, the concentration of any drug treatment must also be optimized for similar reasons. These optimized parameters have been thoroughly investigated in the course of thousands of screens and released as a robust protocol [8].  1.4.2 Methods to assess competitive fitness profiling 1.4.2.1 Microarray The Affymetrix TAG4 array (Genflex tag 16K array) has 5 copies of each feature (sets of complements to each barcode) distributed across the array [7]. The concordance across these features allows for the detection of hybridization effects, and regions dense for outliers can be masked from further analysis. Oligonucleotides present but not assigned to a particular strain barcode are used to estimate non-specific hybridization [7, 8]. The TAG4 array was an improvement on the TAG3 array [5] which had large, single region microarray features (24µM2) 11  for each barcode and therefore was sensitive to positional effects such as staining irregularities or air bubbles. The TAG4 array has five 11µM2 features for each barcode and also added 818 ‘repaired’ tags for which there were discrepancies between the original tag sequence and actual tag sequence as detected by barcode sequencing [7].  1.4.2.2 Barcode sequencing (Bar-seq) Uptags and downtags can also be detected by sequencing by amplifying barcodes using a limited number of PCR cycles with primers that include the Illumina adapter sequence. Bar-seq has improved accuracy and sensitivity over microarrays, because neither non-specific hybridization of probes nor saturation of signal intensity is a concern [9]. The ability to multiplex samples is making Bar-seq a cost-comparable alternative to microarrays for assessing relative strain abundance. To date, only a handful of papers using this method have been published [9-11, 44].   1.5 Thesis rationale While the initial studies with the PBprot and BCprot collections revealed new insights, no systematic, genome-scale comparison with a parent auxotrophic collection exists. To uncover the true differences between auxotrophy and prototrophy in an unbiased manner, a benchmark comparison needs to be made. Knowing that auxotrophy is a confounding factor in many studies, I wanted to know the extent of the impact on experimental phenotypes that would be produced when using an auxotrophic versus prototrophic collection, particularly for the amino acid and nucleotide biosynthetic pathways that have been restored. Given the interests of our lab in using chemogenomics to understand drug-gene interactions, I was also interested in the differences that might be obtained when interrogating drugs that have an impact on metabolic pathways.   12  I took advantage of the unbiased metric of the molecular barcodes to systematically characterize and compare the phenotypic behaviour of the deletion sets by performing comparative chemogenomic and environmental fitness profiling of the PBprot and BCprot collections with the original YKOaux collection. Homozygous deletion profiling (HOP) with the diploid YKOaux collection has been widely used over the haploid collections, as secondary site mutations in the latter will affect the only copy of a gene. Both of the prototrophic collections restoring the MATa collection to prototrophy are haploid, and phenotypes due to secondary site mutations are expected [34]. However, to minimize the likelihood of confounding secondary site mutations in the benchmark collection, the diploid BY4743 background was chosen in this study as the reference collection. This choice also allowed me to leverage the large amount of data from other studies [12, 14, 15, 42] to confirm the phenotypes in my fitness profiles. Therefore, it is possible that some of the phenotypes discussed may be due to differences between the haploid and diploid cellular state, rather than auxotrophy versus prototrophy.  All competitive fitness profiling screens were quantified by microarray hybridization. In an effort to robustly determine the strains present and absent in the auxotrophic and prototrophic collections, a subset of 37 samples were sequenced, as well as hybridized to a microarray using the same genomic DNA for comparison. This was also the starting dataset for our laboratory’s validation for fitness profiling by Bar-seq. Overall 220 microarray chips and 37 barcode sequencing samples were produced for comparison of the three deletion collections in a total of 16 media and 6 drug conditions (Table 2).  I first assessed the set of deletion strains detectable by fitness profiling using microarray hybridization to determine the gene sets available for comparison. I used synthetic dropout media lacking individual amino acids to evaluate discrepancies in amino acid and nucleotide pathways 13  between the auxotrophic and prototrophic collections. Conditions that could not be interrogated in an auxotrophic background (HIS-, LEU-, URA-, MET-) I assessed in both prototrophic backgrounds. I also interrogated drugs that either did or did not impinge on metabolic pathways. I expected to see differences between the auxotrophic and prototrophic collections due to genetic background, and that data comprises part of my thesis. Additionally, I found differences between these collections that could be attributed to their respective construction methodologies. While some of these differences may have been predicted, they were not advertised or reported. Accordingly, it would be difficult to interpret these differences without the benchmark comparison that the YKOaux collection provided in this work.  Comparative analysis of fitness profiles revealed both genotypic (biological) and technical (collection-specific) fitness effects. In the following chapter, I highlight the advantages and disadvantages of each collection for the interrogation of specific biological processes. I will show that screening these three collections in parallel helped to 1) untangle genotype-specific responses and 2) discover unintended effects of the methodology used to restore prototrophy, including significant strain loss and the inability to fully restore wildtype behaviour.  14  Chapter 2: Exploring the impact of prototrophic restoration on genome-wide phenotypes  2.1 Introduction In this chapter I investigate the differences between the auxotrophic (YKOaux) and prototrophic (BCprot, PBprot) collections, beginning with the assayable strain content of each collection. With this prerequisite knowledge in mind, I then discuss the results of challenging these 3 deletion collections in 16 media and 6 drug conditions, a total set of 220 microarray chips from which to determine subtle differences between auxotrophy and prototrophy on the state of the cell. Since barcode sequencing is a more sensitive platform by which to determine strain presence than chip hybridization, I then compared results obtained by microarray with those obtained by Bar-seq for a subset of 37 samples. The results of this study reveal that the construction methods used to restore prototrophy to the YKOaux collection result in collection-specific biology that confounds data interpretation. Therefore significant previously undisclosed caveats are associated with the prototrophic collections which are arguably as problematic as the auxotrophies their creators were trying to correct.  2.2 Results 2.2.1 Genetic roster of each deletion collection To compare the different deletion collections, I needed to determine the baseline number of deletion strains present in each collection. While each collection comes with a list of strains present on 96-well plates, this does not mean that all annotated strains will be present by 15  microarray hybridization. This can be due to poor hybridization by the tags, poor PCR-amplification of the tags from a mutation in the universal primer sequence, unequal starting pool representation due to sampling error, or strains that are lost over a number of experimental generations due to slow growth. First, a pool of all deletion strains was generated for each collection (2.4.1.2). Strain presence for each pooled collection was quantified by microarray signal intensities following five generations of growth in Synthetic Complete (SC) media. Significant differences in the distribution of signal intensities observed for each pool (Kolmogorov-Smirnov P < 0.05, Figure 1) necessitated that background thresholds be determined independently for each pool (2.4.2.1).   Figure 1 ECDFs for mean batch-corrected control arrays  Empirical cumulative distribution functions (ECDFs) were calculated using the mean value from triplicate batch corrected SC controls for each collection. Kolmogorov-Smirnoff tests between each pair of deletion pools were used to estimate the significance of the densities shown (P < 0.05), demonstrating that the signal intensity distribution of the three collections are significantly different.  Following removal of strains with signal intensities below background thresholds, a total of 4,776 gene deletions were present in at least one pool and 77% (3,690) of those were present in 16  all three collections. The original YKOaux collection represented the nonessential yeast genes most comprehensively with 96% (4,594) of deletion strains compared to 89% (4,272) in the PBprot and 82% (3,894) in the BCprot collection (Figure 2).   Figure 2 Deletion strain overlap of the 3 deletion collections Venn diagram depicting deletion strains present in each of the 3 deletion collections in SC media (YKOaux, BCprot, PBprot) compared to the universe of strains present in at least one collection (4776).  To highlight the differences between gene rosters of strains present or absent from each collection, pairwise combinations were evaluated for shared and unique functional enrichments. The most surprising difference was the 882 strains below the limit of detection in the BCprot collection, 428 exclusively and 376 also absent from the PBprot collection. These missing strains were phenotypically enriched for genes required for optimal vegetative growth [15], representing 54% and 28% of the missing strains in the BCprot and PBprot collections, compared to 11% for the YKOaux collection (Figure 3).  17   Figure 3 Relative proportion of slow growing strains present and missing from each collection Pie charts illustrating the proportion of 849 established slow growing deletion strains [15] missing or present in each collection. Relative proportions of missing strains from each collection was comparable to a similar study [45].  Furthermore, the 882 genes missing from the BCprot set included nearly all genes required for amino acid and nucleotide biosynthesis, in addition to genes associated with all aspects of mitochondrial function, including the mitochondrial respiratory chain complex and RNA-processing (Figure 4, Table 3). The loss of these strains were anticipated as the majority are unable to grow in minimal media [15] and would therefore be selected against during strain construction.  18   Figure 4 GO enrichment map of the 882 gene deletion strains absent in the BCprot collection GO enrichment map of the 882 gene deletion strains absent in the BCprot collection relative to the gene universe in Figure 2. Two highly enriched biological processes of interest in this study were amino acid biosynthetic processes (P = 2.35E-05) and mitochondrial RNA metabolic process (P = 7.08E-05). Node colors represent different GO biological processes; the node size is proportional to the number of genes present in each enrichment. The width of each edge is proportional to the degree of gene overlap between GO terms. Only the most significant P-value per related GO biological process is shown (see Table 3).  Nearly 60% (45) of the 76 strains absent only in the YKOaux were never constructed as diploids and included genes required for mating. A small subset of these strains were also absent from the BCprot collection and explicitly required for mating (COA1, GPA1, MSL1, SIR2, SIR3, SIR4, SRV2, STE2, STE4, STE5, STE7, STE11, STE14) [4, 5, 46, 47], reflecting the inability of these strains to survive the mating step during construction  The relative proportions of strains present in SC media shown here were consistent to those following five generations of growth in rich media (YPD) (Figure 5).  19   Figure 5 Relative strain presence in YPD is consistent with that observed in SC media. Venn diagram of strain presence in YPD compared to SC for each collection, demonstrating > 97% agreement between the two conditions. Number of strains overlapping between media conditions for the YKOaux (top left), PBprot (bottom left), BCprot (top right), and the gene universe of strains present in at least one collection (bottom right).  2.2.2 Comparative fitness profiling  Having quantified the strain representation of each collection, I sought to determine the difference in biological outcomes between the prototrophic and auxotrophic deletion strains. Each collection was profiled in competitive fitness assays in diverse stress conditions including 1) twelve nitrogen and nucleotide limiting conditions, 2) the DNA damaging agent cisplatin, 3) compounds shown to affect amino acid or nucleotide biosynthesis (PCID 16001701, PCID 16584271, itraconazole), and 4) mitochondrial stress conditions (the oxidative phosphorylation uncoupling agents FCCP, CCCP and growth in obligate respiratory conditions (YPG)), as well as respective control conditions (SC, YPD, T0) (Table 2).    20  Table 2 Drug and media condition replicates assayed per deletion collection. Control Condition YKOaux BCprot PBprot SC Control 3 3 3 SC ADE- 3 3 3 SC ARG- 3 3 3 SC HIS- 0 3 3 SC LEU- 0 3 3 SC LYS- 3 2 3 SC MET- 3 3 3 SC SER- 3 2 3 SC THR- 3 2 3 SC TRP- 3 3 3 SC URA- 0 3 3 SC MM* 3 4 4 SC MM+Ura 0 1 1 YPD T0 (no growth) 5 5 5 YPD control 11 14 12 YPD cisplatin, DNA x-linking agent 3 4 3 YPD CCCP, protonophore inhibitor of oxidative phosphorylation  3 3 3 YPD FCCP, protonophore inhibitor of oxidative phosphorylation  3 3 3 YPD PCID 16001701, novel quinolone compound,  4 4 4 YPD PCID 16584271, affects purine biosynthesis 3 3 3 YPD itraconazole, antifungal that affects the tryptophan biosynthesis 3 3 3 YPD YPG, obligate respiratory condition 2 4 4 * MM condition for YKOaux was supplemented with histidine (20mg/l), leucine (30mg/l), methionine (20mg/l) and uracil (20mg/l).  21  2.2.2.1 Nutrient limiting conditions Fitness profiles readily identify all genes required in the corresponding biosynthetic pathways when assayed in conditions lacking a specific amino acid, purine or pyrimidine [12]. Both the YKOaux and the PBprot collections recapitulated the established biosynthetic pathways and showed significant similarity (P < 0.05, 2.4.2.5) in adenine, arginine, methionine, lysine and tryptophan dropout screens (Figure 6, Figure 11).   22   Figure 6 Hierarchical clustering of FD scores for YKOaux and PBprot collections in synthetic dropout media Hierarchical clustering of median fitness defects scores across 3 replicates observed for YKOaux(blue) and PBprot (green) deletion strains in five synthetic dropout conditions (ADE-, ARG-, MET-, LYS-, TRP-) (log2 ratio >= 1 in at least one condition). The labelled clusters list the deletion strains with fitness defects in the corresponding media dropout condition.   The conditions that prohibit screening of the YKOaux collection, including histidine, leucine and uracil dropout media, were of the greatest interest because of the paucity of functional annotations for these biosynthetic pathways. In these conditions, expression of the genes carried on the PBprot ARS-CEN vector is explicitly required. Fitness profiling of the PBprot revealed a 23  unique gene signature that described genes required for plasmid and mini-chromosome maintenance (Figure 7).   Figure 7 PBprot fitness signature revealed in conditions requiring expression of the pHLUM plasmid Hierarchical clustering of median fitness defect scores (for all deletion strains with log2 ratio > = 1 in at least one experiment) for the PBprot collection in synthetic dropout media (ADE-, ARG-, HIS-, LEU-, LYS-, MET-, SER-, -THR-, TRP-, URA-). Deletion strains belonging to the histidine-leucine-uracil (HLU) signature are boxed in black. Inset: GO biological process enrichment for the HLU signature. Node colors represent different GO biological processes; node size is proportional to the number of genes in each enrichment. Edge widths in the network represent the fraction of overlap between genes in related GO terms. Only the most significant P-value per related GO biological process is shown (see Table 5). 24  Specifically, the 73 core genes in this histidine-leucine-uracil (HLU) fitness defect signature (Table 4) were enriched for biological processes that exhibited a response to DNA replicative stress including: 1) nuclear division (P = 2.75E-12), 2) regulation of mitotic sister chromatid segregation (P = 2.54E-11), and 3) M-phase of mitotic cell cycle (P = 1.02E-12) (Figure 7, inset, Table 5). Many of these genes were originally identified in classic genetic screens for chromosome instability (CIN) [48], including CLB5, CSM3, CTF4, CTF19, ELG1, IML3, IRC15 (overlaps CTF19), KAR3, MCM21, MRC1, NUP120, SPT2, and TOF1.  The HLU fitness defect signature was not observed in the methionine dropout media for the PBprot deletion strains despite this condition also requiring active expression of the MET15 gene from the ARS-CEN vector (Figure 6, Figure 11). This finding is consistent with methionine’s role in the regulation of cell cycle progression. Insufficient levels of intracellular methionine (and its downstream product, cysteine) signal cell cycle arrest at G1/start [49-51] until a sufficient level of metabolites is reached to allow successful progression. During methionine depletion, cell cycle delay may alleviate the fitness effects observed in the HLU signature. If this interpretation is correct, a similar gene signature would be expected to be observed in any condition that impinges on the histidine, leucine or uracil biosynthetic pathways, as well as the associated perturbed processes including cell division/DNA replication.  To correct for the confounding effect of the HLU signature to allow the identification of genes specifically sensitive in histidine, leucine and uracil dropout screens, fitness scores were re-calculated using the HLU gene signature as the reference condition (2.4.2.2). The resulting fitness profiles following this data transformation clearly identified genes known to be required in these biosynthetic pathways (Figure 8, Figure 9, Figure 10). For HIS- this included the 25  general amino acid control (GAAC)-regulated BAS1 transcription factor required only when purine and histidine are specifically limiting. A fitness defect was also observed for BAS1 in the adenine dropout conditions for all three collections (Figure 8, Figure 11).    Figure 8 The prototrophic collections in HIS- dropout media (+/–HLU signature) Fitness profiles for the PBprot and BCprot collections in HIS- after 5 generations of growth. Fitness defects observed for the PBprot deletion strains in HIS- before (left panel) and after (middle panel) correcting for the HLU signature, revealing histidine-specific effects. Right panel: Fitness defects observed for the BCprot collection in the same condition. Red dashed line indicates the significance threshold of a FD score of 1.0. Subtraction of the 73 genes in the HLU signature removed commonly sensitive deletion strains unrelated to the assay treatment and highlights those deletion strains manifesting condition-specific sensitivity.  The uracil profile identified URA1, URA2, and URA5 required for uracil biosynthesis. URA4 was a member of the common HLU signature and was therefore not identified as specifically being required in the uracil dropout condition (Figure 9).   26   Figure 9 The prototrophic collections in URA- dropout media (+/–HLU signature) Fitness profiles for the PBprot and BCprot collections in URA- after 5 generations of growth. Fitness defects observed for the PBprot deletion strains in URA- before (left panel) and after (middle panel) correcting for the HLU signature, revealing uracil-specific effects. Right panel: Fitness defects observed for the BCprot collection in the same condition. Red dashed line indicates the significance threshold of a FD score of 1.0. Subtraction of the 73 genes in the HLU signature removed commonly sensitive deletion strains unrelated to the assay treatment and highlights those deletion strains manifesting condition-specific sensitivity.   The presence of these auxotrophic strains in the collection is unexpected due to selection on minimal media. I learned that in the PBprot collection [29] transformants were selected first for a single auxotrophy, either histidine or uracil (instead on all four markers carried on the complementing ARS-CEN plasmid simultaneously), and in the second passage on minimal media (M. Ralser personal communication). The ultimate reason for the presence of several strains in the collection that should not survive this selection remains to be determined. Compared to the HIS3 and URA3 auxotrophies, deletion of LEU2 is considered more detrimental to cell physiology; such strains exhibit slower growth rates [29] and decreased rate of survival in starvation conditions [10, 52]. When I profiled the collections in the leucine dropout condition, fitness defects in the PBprot included HTD2 from the mitochondrial fatty acid biosynthetic pathway (FASII), genes involved in protein lipoylation (AIM22, GCV3, LIP2), and LPD1 and 27  PDX1 encoding the mitochondrial dihydrolipoyl dehydrogenase complex. The BCprot leucine profile was less informative with fewer strains exhibiting fitness defects compared to the PBprot profile (Figure 10). The BCprot strains with fitness defects included MPC1, a subunit of the mitochondrial pyruvate carrier (MPC1/MPC2), in addition to PDX1, the transcription regulator LEU3, and alpha-isopropylmalate synthase (LEU4), also found in the PBprot profile.   Figure 10 The prototrophic collections in LEU- dropout media (+/–HLU signature) Fitness profiles for the PBprot and BCprot collections in LEU- after 5 generations of growth. Fitness defects observed for the PBprot deletion strains in LEU- before (left panel) and after (middle panel) correcting for the HLU signature, revealing leucine-specific effects. Right panel: Fitness defects observed for the BCprot collection in the same condition. Red dashed line indicates the significance threshold of a FD score of 1.0. Subtraction of the 73 genes in the HLU signature removed commonly sensitive deletion strains unrelated to the assay treatment and highlights those deletion strains manifesting condition-specific sensitivity.  These genes identified in the PBprot screen are known to regulate genes beyond the core leucine biosynthetic pathway. This observation is of particular interest because leucine is thought to play a role in central metabolism, including iron-sulfur cluster biogenesis, mitochondrial genome maintenance, and regulation of acetyl-CoA between mitochondrial and cytoplasmic compartments [53]. The FASII pathway, highlighted in the PBprot profile, is thought to provide 28  the octanoic acid required for biosynthesis of the cofactor lipoic acid (AIM22, GCV3, LIP2), which in turn is required by the mitochondrial pyruvate dehydrogenase complex (LPD1, PDX1) and suggests the leucine biosynthetic pathway plays a substantial role in maintaining healthy mitochondrial function [54]. These observations on the leucine pathway, which cannot be assayed with the YKOaux collection, highlight the usefulness of the PBprot collection. Because the majority of deletion strains required for amino acid and nucleotide biosynthesis are missing from the BCprot collection, the resulting screens in dropout media reported fitness defects only for genes that enhance growth in the absence of particular metabolites, including several of the general control of amino acid synthesis (GCN) system including, ARO3, ARO4, GCN3 and GCN20 (Figure 11). GCN4 however, was present only in the YKOaux collection due to the requirement for this important transcriptional activator during the nutrient selection steps required for the construction of both prototrophic collections [55].  29    Figure 11 Microarray fitness profiles for 4 synthetic dropout conditions Collection-specific fitness profiles representing the median FD of 4 control replicates (SC) to 3 treatment replicates for the YKOaux, PBprot, and BCprot deletion collections in (a) ADE-, (b) LYS- (c) MET- and (d) TRP-.  30  Despite the loss of many strains in the BCprot collection, it is important to point out that by virtue of the backcrossing to a strain prototrophic for the YKO auxotrophies, deletion strains carrying secondary site mutations may have been rescued during the mating step of crossing the MATa collection to a MATα prototrophic strain. To test this hypothesis, fitness defects were found that were observed in both PBprot and YKOaux deletion strains in mitochondrial stress conditions and were not observed to have fitness defects for the BCprot deletion strains. Using these stringent criteria 12 such strains (ABF2, DCS1, ERG3, GIN4, LCB5, MAC1, MRP51, POS5, QRI7, RML2, and TUF1) were identified (Figure 12).  31   Figure 12 PBprot and YKOaux, but not BCprot deletion strains exhibit fitness defects under mitochondrial stress Three conditions that perturb the mitochondria (a) CCCP and (b) FCCP) or (c) rich media with glycerol as the carbon source (YPG), produced similar responses for the YKOaux and PBprot deletion strains, while in contrast, the same 12 deletion strains were largely unaffected in the BCprot collection. 32  All 12 of these genes are annotated as being unable to grow in obligate respiratory conditions [21]. It is therefore possible that there may be secondary site mutations present in these deletion strains that were rescued in the BCprot collection.  In the middle panel of Figure 12c, where YPG is the treatment condition, the YKOaux deletion strains do not recapitulate the same fitness defects seen under proton ionophore stress (CCCP, FCCP). Normalization of YPD control microarray chips together with YPG treatment chips is not ideal as five generations of growth do not end concurrently, and the distribution of the signal intensities for these conditions is different, as more deletion strains exhibit fitness defects in a non-fermentable carbon source. Additionally, this panel profile was the result of two rather than three YPG replicates (as for all other panels). It may be preferable to evaluate the YPG condition relative to a second non-fermentable carbon source, such as acetate, or a second perturbation in YPG, which would provide a similar distribution and endpoint for normalization. However, using the same criteria for the minimal media dropout conditions, a set of BCprot deletion strains was found that exhibited no fitness defects yet are required for lysine (APT1, LYS2) and arginine (ARG5,6, ARG8, ORT1) biosynthesis and moreover, are synthetic lethal with the CAN1 and LYP2 deletions in the BCprot genetic background. These results make it difficult to distinguish potential mutant rescue from mutations that may have been introduced during strain construction. Taken together, these results serve to illustrate that during the construction of any collection, secondary site mutations will arise. However, the inclusion of selections steps may amplify the frequency of such strains by imposing strong selective pressure.  33  2.2.2.1.1 Genotype-specific phenotypic behaviour of the cpa1∆ strain A unique BCprot fitness defect was observed for the strain deleted for CPA1, a gene required for arginine biosynthesis. The BCprot cpa1∆ strain exhibited severe fitness defects in all conditions except in minimal and uracil dropout media (Figure 13, Figure 16).   Figure 13 Collection-specific phenotypes of the cpa1∆ strain across synthetic media conditions Fitness defects are shown for the cpa1∆ strain across all three deletion collections for different synthetic media. The PBprot and YKOaux cpa1∆ strains exhibit fitness defects only in media containing uracil and lacking arginine. The BCprot cpa1∆ strain, in contrast, was only fit in media lacking uracil.   In contrast the YKOaux cpa1∆ strain was sensitive only in arginine dropout media, while the PBprot cpa1∆ strain was sensitive in arginine dropout media, as well as minimal media with the addition of uracil (unassayable in the YKOaux collection) (Figure 13, Figure 14, Figure 16).  34   Figure 14 Competitive fitness profiling of 3 deletion collections in synthetic media lacking arginine Fitness profiles in ARG- for the YKOaux, PBprot, and BCprot collections. Orange labels: Fitness defects in genes known to be required for arginine biosynthesis. Red dashed line indicates the FD threshold of 1.0.  Because the deletion of CAN1 (encoding the major arginine transporter) is synthetically lethal with genes in the arginine pathway [37], I investigated how the cpa1∆ strain was able to survive selection on minimal media during construction of the BCprot collection and why it manifested such an unusual phenotype. An explanation was provided by classical biochemical studies examining the carbamoyl phosphate pathway [56] where both CPA1 and URA2 encode carbamoyl-phosphate synthase (CPS) (Figure 15, -uracil). The activity of either is sufficient for supplying carbamoyl phosphate, the metabolic product required for arginine and pyrimidine biosynthesis. In minimal media, URA2 is expressed and active, allowing the cpa1∆ strain to grow normally. Addition of uracil to the media represses the URA2 enzyme through negative feedback, and the cpa1∆ strain is inviable due to the absence of any carbamoyl-phosphate synthetase activity (Figure 15, +uracil).  35    Figure 15 Crosstalk between the arginine and pyrimidine metabolic pathways via carbamoyl phosphate Schematic depiction of the crosstalk between the arginine and pyrimidine biosynthetic pathways via the metabolite carbamoyl phosphate. In the absence of arginine and uracil supplementation (red) the cpa1∆ strain grows normally by “borrowing” carbamoyl phosphate produced by the pyrimidine biosynthetic pathway. If uracil is added in these conditions (green), the pyrimidine pathway is repressed, and growth of the cpa1∆ strain is prohibited, phenocopying the synthetic lethal interaction of ura2∆ with cpa1∆.  Viability can be rescued by adding arginine to the media as evidenced by the PBprot cpa1∆ strain in synthetic complete or minimal media with added uracil and arginine (Figure 16). However, the BCprot cpa1∆strain is also can1∆ (the deletion of the major arginine transporter), which prohibits rescue even in the presence of excess arginine (Figure 16).  36   Figure 16 Rescue of the cpa1∆ strain with arginine supplementation is dependent on genetic background Collection-specific phenotypes observed for the cpa1∆ strain by individual strain analysis. The PBprot and BCprot cpa1∆ strains grow normally in minimal media (MM), and neither grows in the presence of exogenous uracil (MM +Ura). Arginine amounts present in standard SC media (2mg/ml) is sufficient to rescue the PBprot cpa1∆ phenotype (SC), as is true for the YKOaux cpa1∆ strain (data not shown). In contrast, even arginine concentrations 6.7x higher (13.3mg/ml) are insufficient to fully rescue the BCprot cpa1∆ strain due to the deletion in CAN1 (encoding the major arginine transporter) present in the BCprot genetic background (MM+Ura+Arg).  This can1∆ cpa1∆negative genetic interaction therefore explains the fitness defects observed for the BCprot cpa1∆ strain in all conditions where uracil is present. Interestingly, data mining of Robinson et al. (2014) showed this CPA1 fitness defect was also observed in the BCprot when grown in rich media with dextrose as the carbon source (YPD) but not in rich media with galactose as the carbon source (YPGal). In light of our measurements, we can explain this observation by the fact that YPGal metabolism requires uracil, which would activate the pyrimidine biosynthesis pathway, initiate carbamoyl phosphate production, and relieve the fitness defect of the cpa1∆ strain.  37  2.2.2.2 Drug and small molecule stress conditions Results from chemogenomic profiling of the DNA cross-linking agent cisplatin for all three collections were consistent with established mechanisms and previous genome-wide fitness studies [42, 57, 58]. Strains exhibiting drug sensitivity in all three collections were significantly enriched for specific DNA damage response (DDR) processes that included, for example, nucleotide excision repair (NER) (RAD1, RAD2, RAD4, RAD10, RAD14), homologous recombination repair (HRR) (RAD51, RAD55, RAD59), post-replication repair (PRR) (RAD5, RAD18), translesion synthesis (TLS) (REV1, REV3) and PSO2, which is required for repair of cisplatin-induced interstrand cross-links (Figure 17).  Figure 17 Competitive fitness profiling of 3 deletion collections in cisplatin Fitness profiles of the median fitness defect scores across 3 replicates for the YKOaux, PBprot and BCprot collections in the DNA cross-linking agent cisplatin. Red labels: Fitness defects observed for strains deleted for genes annotated as being involved in DDR (DNA damage response). Red dashed line indicates a FD score of 1.5.  Interestingly HIS5 appears with a significant fitness defect in the PBprot profile, but not in the YKOaux or BCprot profile (Figure 17). Since there is crosstalk between the histidine and adenine 38  pathways [59], it is possible that there is an impact on histidine biosynthesis during DDR. In addition, of the 36 genes involved in DDR with a significant fitness score in at least one collection, ~60% (22) exhibit slow growth only in YPD but not in minimal media [15]. Reflecting the ability of the PBprot and BCprot collections to maintain these strains during collection construction in minimal media, ~75% of these genes were present in those two collections, compared to ~25% in the YKOaux collection.  S. cerevisiae has greatly contributed to the understanding of mitochondrial function, due to the dispensability of its mitochondrial genome and the ability to switch metabolic states based on the carbon source. In the yeast, approximately 1,000 of all 6,000 genes participate in mitochondrial processes and serve crucial, evolutionarily conserved cellular functions. I therefore focused on conditions that perturb mitochondrial function to compare deletion collections. Fitness profiles of the three collections in low doses of the mitochondrial membrane potential poisons, CCCP and FCCP, as well as growth in obligate respiratory media (YPG, where glycerol provides the carbon source) exhibited strong enrichment in both the YKOaux and PBprot collections for mitochondrial translation and respiration (P < 1E-17 in all three conditions).  39   Figure 18 Competitive fitness profiling of 3 deletion collections in FCCP Fitness profiles of the median fitness defect scores across 3 replicates for the YKOaux, PBprot and BCprot collections in the proton ionophore FCCP. Green labels: Fitness defects observed for strains deleted for genes annotated as being involved in mitochondrial processes. Red dashed line indicates a FD score of 1.5.  In contrast, the BCprot fitness response was relatively sparse; no enrichment was observed in any of the mitochondrial stress conditions, reflecting the significant proportion of mitochondrial deletion strains missing in that collection (Figure 18).  Challenging the deletion collections with a compound of unknown mechanism provides an unbiased stress for comparing the three collections. The fitness signature of a cationic quinolone (PCID 16001701) previously screened by our laboratory [14] resembled the adenine dropout fitness profiles (Figure 11), suggesting the compound acts via a mechanism that (directly or indirectly) requires adenine biosynthesis. Rescreening of this compound reproduced this gene signature in YKOaux and was supported by similar profile in PBprot, identifying adenine and folic acid (a cofactor required in for several steps of adenine biosynthesis) biosynthetic genes (ADE1, ADE3, ADE4, ADE6, SHM2, THI3) (Figure 19).  40   Figure 19 Competitive fitness profiling of 3 deletion collections in a cationic quinolone Fitness profiles of the median fitness defect scores across 3 replicates for the YKOaux, PBprot and BCprot collections in a novel quinolone compound (PCID 16001701) [14]. Blue labels: Fitness defects for strains deleted for genes annotated as being involved in adenine biosynthesis. Red dashed line indicates a FD score of 1.5.  The BCprot profile was uninformative with respect to the mechanism of action of this particular compound.  The trend that the YKOaux and PBprot share similar fitness profiles for drugs that impinge on amino acid or nucleotide biosynthetic pathways while the BCprot profile was limited in its usefulness due to missing strain content, was also seen for screens in PCID 16584271 and itraconazole (these profiles included sensitive strains whose gene deletion impacts adenine and tryptophan biosynthesis, respectively). Accordingly, I would not recommend the BCprot collection for investigation of metabolic pathways or unknown compound mechanisms. 2.2.3 Bar-seq validation of microarray data Sequencing of the molecular barcodes (Bar-seq) for select samples from the same genomic DNA that was assessed by microarray provided further validation of strain presence. Strains for which 41  all counts were greater than or equal to 50 across four SC replicates were considered to be present. By both methods, 4,535 strains were considered to be robustly present for the YKOaux collection. A unique set of 160 strains had counts above background by Bar-seq, and 59 strains were detected by microarray only. The microarray intensity distribution of strains present only by one method was similar when both uptags and downtags were considered (Figure 20). Further investigation revealed 938 barcodes with one or more mismatches from the annotated barcode [8], as has been previously described [9]. For the subset of strains present only by microarray 55 out of 59 strains (93%) had one or more tags with a mismatch, compared to 19 of the 160 strains (12%) present only by Bar-seq, and 635 of the 4535 strains (14%) present by both platforms. One possible explanation for strains detected by microarray but not by Bar-seq is that mismatched tags have non-specifically hybridized to the array, giving a false positive signal. 42   Figure 20 Strain presence assessed by microarray is consistent, but less sensitive compared with Bar-seq Signal intensity and count density distributions for the 4,754 strains called present by at least one method in the YKOaux collection. The distribution of a) counts for strains detected as present or missing by Bar-seq, or present by both microarray and Bar-seq, and (b) array intensities for strains detected as present or missing by microarray, or present by both microarray and Bar-seq. The separation of signal-to-noise in the microarray data was poor for strains for which the two methods were not in agreement compared to the Bar-seq data. 43  Replicates from three synthetic dropout conditions (ARG- LYS-, TRP-) were also evaluated by both methods (Figure 21). Comparison of the ordered gene lists generated from Bar-seq and microarray data shows significant similarity that was not due to chance (B = 1000000, P < 0.05) for all compared media conditions (2.4.2.5). Thus, Bar-seq and microarray hybridization were sufficiently reliable to detect the top strains sensitive to the treatment conditions tested. 44   Figure 21 Microarray fitness profiles were independently confirmed by Bar-seq The fitness defects obtained via barcode sequencing (Bar-seq) (left) and microarray (right) in synthetic dropout conditions for (a) ARG-, (b) LYS-, (c) and TRP-. Comparison of ordered gene lists for each condition-method pair showed significant similarity (P < 0.05).  45  2.3 Discussion I have shown that a comparative, genome-wide fitness survey of the original YKOaux collection and two prototrophic derivatives across diverse environmental and stress conditions revealed several surprising findings relevant to applying these collections in gene functional studies. First, while both the PBprot and BCprot collections satisfy the definition of prototrophy: ‘that a cell or organism has the same nutritional requirements as wildtype’, the benefits of prototrophy are offset by the cost of losing informative deletion strains. For example, the selection on minimal media during the construction of the prototrophic collections by definition prohibit future study in these basic nutrient conditions, as informative strains unable to grow will be selected against. This study demonstrates that non-neutral selections during strain construction introduced unpredictable genome-wide bias, as observed for the BCprot and PBprot collections, and in many respects these biases were far more constraining than the YKOaux collection’s nutritional auxotrophies. While the repair of the YKO auxotrophies by genetic complementation in PBprot collection was more effective than the BCprot collection (with respect to strain loss) it was not neutral, and certainly not phenotypically equivalent to wildtype in conditions requiring expression from the ARS-CEN vector. Phenotypic differences between episomal and integrated genetic complementation are well documented [60, 61]. I therefore expect that, despite the ability to successfully correct for a well-defined HLU signature, unanticipated episomal effects are likely to occur that will escape detection. In contrast, though the BCprot collection restored prototrophic markers to their native location (with the exception of the HIS3 ortholog from S. pombe from a non-HIS3 promoter), the CAN1 and LYP1 deletions present in the genetic background also introduced biases, as demonstrated for the cpa1∆ can1∆ synthetic lethal phenotype. These construction-specific effects also disrupt a key feature of competitive fitness 46  assays – namely that the relative strain abundance in the starting pool be approximately equal. Nutritional selection steps skew this initial distribution, particularly when multiple strain passages are part of the construction methodology. As a result, the ability to detect fitness defects becomes more difficult, as reflected by the higher background thresholds and signal intensity distributions of the prototrophic collections, compared to the YKOaux collection.  The unexpected liabilities present in the prototrophic collections underscore that highly engineered versions of YKOaux collection are more constrained than generally assumed. Informative strains lost from the BCprot collection (~900 strains) share significant biological enrichment for genes involved in amino acid and nucleotide biosynthesis as well as mitochondrial processes that compromise the ability to interrogate these processes, and limit the biological space the collection was created to expand. The ability to perform such a precisely genetically controlled study on three genotypically distinct deletion collections in S. cerevisiae is not currently feasible in other systems. These results therefore may provide insight into fundamental principles of genotype-by-environment relationships and suggest that condition-dependent cellular responses (i.e. phenotype) are greatly influenced by genotype.  The experimental design and assay constraints described here may help guide screens in other organisms and cell lines as they become tractable using CRISPR and other genome-editing techniques. The importance of systematically benchmarking genomic libraries will be critical to establish and maintain the quality of functional and phenotypic gene annotations. Finally, I hope this study will serve to encourage and guide the design of future yeast deletion collections, most 47  notably, the need to move beyond derivative YKOaux libraries to the de novo construction of a prototrophic collection.  2.4 Materials and methods 2.4.1 Experimental methods 2.4.1.1 Yeast deletion strains and media preparation The YKOaux deletion collection is from the original stock center of the Saccharomyces Genome Deletion Project [4], curated and maintained by Angela Chu at the Stanford Genome and Technology Center. The PBprot [29] and BCprot [20] deletion collections were kindly provided by the Ralser and Caudy laboratories. Synthetic complete and amino acid dropout media was purchased from Sunrise Science Products. 2.4.1.2 Construction of gene deletion collection pools The diploid YKOaux collection and the haploid BCprot and PBprot collections were pinned (S & P robotics Inc., BM3-BC) from thawed glycerol stocks in 384-well or 96-well plates respectively onto rich YPD media (20 g/L bacto peptone, 10 g/L yeast extract, 20 g/L bacto agar and 20 g/L glucose), and recovered for 48-72h at 30 °C until colonies reached 2mm in diameter. Plates were flooded with 12 ml liquid media and yeast cells were soaked and scraped off the plates. Resuspended cells from each plate were pooled in a sterile flask, and the final OD600 of the pool was adjusted to a final 50 OD600 ml-1. DMSO was added to the pool to a final concentration of 7% (v/v), mixed well, aliquoted into individually capped PCR tubes and stored at -80 °C.   48  2.4.1.3 Competitive fitness profiling 2.4.1.3.1 Synthetic media HOP screens 700μl of pooled aliquots at a starting OD600 of 0.0625 were grown in duplicate wells on the same plate for five doublings in a Tecan Genios (Tecan Systems, Inc.) spectrophotometer at 30 °C. Pellets were manually harvested (synthetic media and drug screens) or automatically collected (YPG screens) using a Packard Multiprobe (Perkin Elmer) liquid handler and stored at -20 ºC in individual tubes after pellet washing (drug screens) for no longer than 1 day. For the amino acid drop-out experiments, each pooled collection was grown in synthetic complete medium (SC, Sunrise Science Products) or rich media (YPD) as the control conditions, and SC media with an individual amino acid of interest dropped out as the treatment condition. For the YPG experiments, 3% glycerol was the treatment condition and YPD was the control condition 2.4.1.3.2  Chemical HOP screens Samples were subject to the same starting OD600 and doubling times as above. Screening concentrations for each compound (cisplatin (Toronto Research Chemicals), PCID 16001701 (ChemDiv), CCCP (Sigma), FCCP (Sigma), itraconazole (Toronto Research Chemicals), PCID 16584271 (Vitas-M Laboratory)) were determined by analyzing dose-response curves on wildtype auxotrophic BY4743 or prototrophic BY4743 with pHLUM to determine the concentration that inhibits BY4743 growth by 15 – 20% (200μM for cisplatin, 125μM for PCID 16001701, 32μM for CCCP, 6μM for FCCP, 0.8μM itraconazole, and 125μM PCID 16584271). YPD plus solvent (2% DMSO) was used as the control with the exception of cisplatin (2% H20).   49  2.4.1.3.3 Genomic DNA extraction and hybridization to an Affymetrix TAG4 microarray Following growth, genomic DNA was extracted from cell pellets using YeastStar Genomic DNA Kit (Zymo Research, catalog #D2002) and quantified using the NanoDrop 2000 (Thermo Scientific). Uptag and downtags were amplified separately, pooled, and hybridized to Affymetrix TAG4 microarray (Genflex tag 16K array v2) as previously described [62]. 2.4.1.4 cpa1∆ strain phenotype validation The BCprot and PBprot cpa1∆ strains were grown individually from a starting OD600 of 0.0625 to saturation in minimal media (MM), SC, MM + uracil (20µM), and MM + uracil (20µM) + arginine (76µM) as shown in Figure 16 or as described for pooled growth. 2.4.1.5 Library preparation Barcode sequencing libraries were prepared using a custom two-step PCR approach using Phusion High-Fidelity DNA Polymerase (Thermo Fisher). First, uptags and downtags were separately amplified as described above for competitive fitness assays, but using primers pairs UP_F TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGATGTCCACGAGGTCTCT and UP_R GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGTCGACCTGCAGCGTACG or DOWN_F TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGAAAACGAGCTCGAATTCATCG and DOWN_R GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGCGGTGTCGGTCTCGTAG. Uptag and downtag PCRs were then pooled in equal amounts and purified using the GeneJET PCR 50  purification kit according to manufacturer’s instructions (Thermo Fisher). Second, purified barcodes were diluted 1:10 and 1µl used as template in the second PCR using Nextera XT index primers (Illumina), which contain individual barcodes as well as Illumina adapters. Cycling conditions were as follows: 98 °C for 30 s; 8 cycles of 98 °C for 10 s, 55 °C for 30 s, 72 °C for 15 s; 72 °C for 5 min. Libraries were then purified using Agencourt AMPure XP beads (Beckman Coulter) at a ratio of 3:5 beads to DNA, checked on Agilent High Sensitivity DNA chips for the Bioanalyzer (Agilent), and quantified using Quant-iT high sensitivity dsDNA Assay kit (Thermo Fisher). Pooled sequencing libraries were sequenced on a HiSeq 2500 (Illumina) in rapid run mode, generating paired or single-end 100 bp reads. 2.4.2 Data analysis 2.4.2.1 Array normalization and pre-processing Each probe on the TAG4 barcode microarray (Genflex tag 16K array v2, Affymetrix) is represented by five replicate features dispersed across the array that allow hybridization artifacts to be identified and corrected. Hybridization artifacts were removed using a previously described masking algorithm [8]. Independent sample sets were defined by collection and growth media (9 sets in total, PBprot, BCprot and YKOaux in SC, YPD and T0). I used a two-component Gaussian mixture model to fit the distribution of tags in the control arrays in each set (R version 3.2.2, mixtools package, version 1.0.4) [63]. The estimated components represent tags that successfully hybridized (present) and tags that did not successfully hybridize (absent). I used the posterior distribution of the assignment of a tag to the present or absent component to select present tags for further analysis in the treatment conditions. To be called as present, all tags representing an ORF had to have a posterior value greater than 0.5 in all of the control replicates. During the 51  course of this study I recognized that the homozygous BY4743 pools I used in this study were missing a subset of 143 strains due to a technical error that occurred during pool construction. These strains were added to the collection in the YKOaux collection version 2.0 (http://www-sequence.stanford.edu/group/yeast_deletion_project/ykov2.html) and were present in both the BCprot and PBprot collections. For the sole purpose of the venn diagram in Figure 2, the homozygous strains in this subset were counted as present using the data from > 3000 experiments previously generated in our lab in a large-scale chemogenomic study [14].  Next, tags designated as present upstream (uptags) and downstream (downtags) of the drug resistance cassette were normalized to the overall median across arrays within each set. The uptag and downtag for each ORF were collapsed into a single value by selecting the ‘best’ tag; the tag that exhibited the lowest coefficient of variation across the control replicates for each set. Biological replicates for each condition were performed in triplicate and batch corrected for technical variation using the Combat function in the R sva package (version 3.14.0) [64]. 2.4.2.2 Fitness defect scores Fitness defect scores (FD) for each tag (representing a deletion strain) in each set were calculated by subtracting the log2 intensity value of the treatment condition from the corresponding median control value of each individual tag. To estimate strains exhibiting significant fitness defect scores, triplicate replicate values for each condition were fit to a linear model. The resulting P values were corrected using the Benjamini-Hochberg method to adjust for multiple comparisons [62]. The PBprot HLU signature was comprised of genes that had a FD > 1 across the medians of all HIS-, URA- and LEU- replicates. The PBprot HLU signature was corrected by subtracting out 52  the median FD score for each deletion strain across HIS-, LEU-, and URA- synthetic dropout conditions.  2.4.2.3 Gene Ontology (GO) enrichment analysis GO enrichment analysis was performed in Cytoscape (version 3.3) [65] with the ClueGO plugin (version 2.2.5) [66]. A gene set of interest was compared to the custom gene universe of the set of strains that were present in at least one of the three deletion collections. A minimum of five genes in the gene set was necessary to be associated with a GO Biological Process term. A right-sided hypergeometric test was used with a Bonferroni step-down correction and a minimum P value of 0.0005 with a kappa score threshold of 0.4 [67]. Node sizes shown in the figures were proportional to the number of genes found in the gene set associated with the term. 2.4.2.4 Bar-seq analysis For the Bar-seq libraries sequenced with a paired-read protocol the read mates were merged into single reads using BBMerge version 8.82 from the BBTools/BBMap analysis suite (https://sourceforge.net/projects/bbmap/). Following that preliminary step, the same analysis procedure was then used on the reads originating from all the libraries, sequenced with paired or single read protocols. Briefly, Bar-seq single sequence reads were first trimmed to 50 bases with Trimmomatic version 0.33 [68] and then mapped to a yeast barcode database using the short-read aligner BWA version 0.7.12 [69]. The BWA database was built using the barcode information from Pierce et al. (2007) with the concatenation of barcode primer sequences at both ends of the barcodes specific for the uptags and downtags. Filtering of the aligned reads was performed with the SAMtools toolbox version 1.2 [70], keeping only reads with mapping quality of 30 and above. Reads were counted for each library with the help of the BEDTools suite 53  version 2.24 [71] and a matrix of counts was created for the whole dataset with a custom-made Perl script for downstream statistical analysis. After filtering for tags that had >= 50 counts across all control replicates, uptags and downtags for each strain were summed, normalized and analyzed with the edgeR package version 3.10.5 [72] as previously described [11]. 2.4.2.5 Similarity of ordered gene sets The similarity of gene lists, ordered by fitness defect scores for common conditions (ADE-, ARG-, LYS-, MET-, TRP-), was compared for the YKOaux and PBprot collections using the OrderedList package version 1.40.0 [73]. An ordered gene set comparison was also performed to compare the Bar-seq and microarray data for the YKOaux collection (ARG-, LYS-, TRP-). The overlap of the gene lists were calculated and weighted towards the 30, 50, and 100 highest genes in the order, recognizing that the majority of genes do not change significantly between the SC control and the synthetic dropout conditions. Lists were subsampled one million times (B=1000000) to generate a distribution of random similarity scores, to which the observed similarity score was compared. Even at the highest weighting (the number of genes with a FD>=1), all samples tested were statistically similar with P < 0.05.  54  Chapter 3: Conclusion 3.1 Summary While construction of the YKO collections was a multi-laboratory effort which took 5 years and cost over $700,000 USD in 1997 (equal to $1.1 million USD today), these prototrophic collections were each a single laboratory’s attempt to find a solution with the resources at hand. Research is often about balancing the caveats of experimental methods, cost, and design. As researchers we must be cognizant of these caveats when evaluating large-scale projects, and control for these constraints to the best of our ability. Nonetheless, any compromises must not interfere with the scientific question that one is asking with a particular technology. In other words, while a technology may not be perfect, it has to be suitable. Therefore, in order to define the scope of applications accessible with these prototrophic deletion collections in contrast with the original auxotrophic collection, elucidating the interactions between multiple auxotrophic markers, and reporting on collection-specific biology is extremely important. Beyond defining the performance of these collections, with this research I have shown that the ability to resolve subtle but important metabolic differences in a genome-wide manner was confounded by the construction-specific artifacts of the deletion collections being compared. Furthermore these biasing effects would not have been apparent (as evidenced by the fact that they were not reported in other studies) if I had not be able to directly contrast them with the benchmark of the YKOaux collection.  This work serves to show that ‘repairing’ an auxotrophic deletion collection, whether by gene integration or plasmid complementation, creates additional confounders that were not present in the parent collection. It may be possible to control for these biological artifacts as with subtraction of the HLU signature for the PBprot collection, or impossible in the case of the 55  missing >800 strains for the BCprot collection. Even within these constraints, there may be applications for each collection if one is cognizant of the biological caveats. The BCprot collection should not be used for amino acid or mitochondrial studies in general. However, a unique phenotype was found for the cpa1∆ can1∆ strain, prohibiting rescue with arginine in contrast to PBprot cpa1∆ strain, which highlights the benefit of interrogating different genetic backgrounds. The PBprot collection was successfully assayed in conditions prohibited by the YKOaux genetic background, and showed the impacts of replicative stress in conditions requiring gene expression from a plasmid. Screens with this collection further highlighted the difference in the response to cellular stress and cell cycle arrest for the MET- versus HLU- conditions. One successful outcome of the publication of this work would be that the yeast community will recognize both construction-specific and collection-specific caveats associated with the BCprot and PBprot collections, such that data interpretation will not be confounded. This work presents a strong empirical argument for the generation (rather than repair) of a prototrophic seamless deletion collection as well as interrogation of diverse genetic backgrounds. While the auxotrophies of the YKO collection have introduced bias on gene functional annotation, the S288c strain background, and its domestication to a laboratory setting, has also introduced bias into yeast gene annotation, especially with respect to mitochondrial gene phenotypes. A substitution mutation in the mitochondrial polymerase, MIP1, increases mitochondrial DNA instability thereby increasing petite frequency in S288c [74, 75], and a Ty insertion in the transcriptional regulator HAP1, affects many genes involved in electron-transfer reactions [76]. Other biological quirks in this laboratory version of S288c include a nonsense mutation in FLO8, preventing filamentous growth [77], as well as lacking the galactose permease, GAL2, and delayed galactose induction in general compared to other Saccharomyces species [78]. 56  Expanding the yeast biological toolkit with other strains and isolates will reveal the extent of this bias in due time. 3.2 Future directions Researchers are beginning to untangle the epistatic effects of auxotrophies on cell physiology. Strains and plasmid sets have been created with all possible factorial combinations of the BY series auxotrophic markers that can be used systematically to study the effect of each marker and related-combinations on growth rate, gene expression, protein production and metabolic profiles [29, 79]. In addition to the finding that growth differences between these strains was largely driven by the LEU2 deletion with unpredictable epistatic effects [29], transcriptome profiling also revealed metabolic-genetic interactions that showed 85% of expressed genes were affected by at least one auxotrophy [79]. Transcriptome profiling data proved to be highly susceptible to different metabolic-genetic backgrounds; an auxotrophic deletion will have profound effects on cell physiology regardless of background and produces significantly distinct expression results from the parent strain [79]. Indeed the magnitude of transcriptional effects suggests that results would be context-dependent even among the BY series (MATa and MATα having three common and one distinct auxotrophic marker each). As the list of side-effects of the genetic auxotrophies lengthens, it is apparent that fair comparisons between deletion mutants and cross-laboratory reproducibility are challenging and highly-dependent on proper control strains.  While the ‘repaired’ prototrophic collections have their own caveats, they enabled interesting studies that were prohibited in the auxotrophic background. For example; the PBprot collection was used to look at the cross-feeding behaviour and sharing of metabolites at a community level [80]; while another study produced profiles of amino acid metabolites for all deletion strains in 57  the PBprot collection and showed that over 1,000 nonessential genes contribute to the metabolic signature of a cell [35]. The BCprot collection was evaluated in 28 basic carbon and nitrogen environments [39], which contributed additional phenotype information for a subset of uncharacterized ORFs. However, the number of missing strains alone points out the limitation of current prototrophic collections and the need for an improved deletion set.  In addition to bias introduced by auxotrophic markers, the majority of dominant markers have the constraint that they cannot be simply recycled, i.e. there are no easily counter-selectable dominant drug markers. While solutions have been presented, such as the counter-selectable amdSYM dominant marker [81], or dominant marker removal strategies such as the LoxP-Cre recombinase system [82-84] or MX4blaster cassette [85], these tools have been variably adopted, and most leave a genomic ‘scar’ at the site of excision.  While it is necessary to allow for facile genetic manipulations (ie. selection and counter-selection) it is well worth considering ways to ameliorate previously recognized biases when designing a new deletion collection. For example, a plasmid is available that, akin to the PBprot collection’s MATa restoration, could transform the MATα collection to prototrophy [86]. It is difficult to imagine that time and resources will be available for such a basic research undertaking.  Creative applications of technology are providing new ways to create custom deletion strains. For example, a polyploid industrial strain of Saccharomyces cerevisiae, ATCC 4124, was modified by CRISPR-Cas9 technology to introduce auxotrophies simultaneously at all alleles; the efficiency of which had been previously been too low to be useful in a diploid background [87]. While this technology was used to generate auxotrophies in a parent strain, one can imagine applications for constructing prototrophic single-deletion strains. 58  New methods of high efficiency genome editing that maintain a prototrophic background include the Haploid Engineering and Replacement Protocol (HERP) cassettes [88]. The cassettes consist of a Thymidine kinase (TK) dominant marker, coupled to an endonuclease which creates double-strand breaks to increase homologous recombination efficiency. This was reported to be an order of magnitude more efficient than the CRISPR-Cas methodology recently reported [87, 88]. HERP cassettes were used to delete the ADE2 coding sequence in seven Saccharomyces species, as well as providing successful deletions in both haploid and diploid genetic backgrounds. HERP cassettes have not yet been used at the genome-scale [88], perhaps due to the initial genetic modification needed to make the HERP platform useful. Lastly, in addition to the auxotrophies present in the YKO collection, variants in the S288c parent strain are associated with genotype-specific traits including poor sporulation and increased rates of mitochondrial genome loss [89-92], neither of which reflect wild yeast traits. Furthermore, within the S. cerevisiae clade, S288c has been classified as an outlier both at the sequence level and by comparative phenotyping [1, 93]. Within S. cerevisiae populations, genome content variation was found to be higher than within populations of its phylogenic neighbour S. paradoxus [94]. However, even a closely related S. cerevisiae strain used in the laboratory, ∑1278b, shows different phenotypes than S288c for 10% of gene deletions [95].  In light of these observations and caveats, no single deletion collection will serve all experimental needs, and compromises will be made each time a new collection is created. I would therefore advocate for the development and assaying of more divergent yeast collections such as those described for pseudo-filamentous or enological strains [95, 96] and closely related 59  human pathogens [97] as sequencing data becomes available and methods for seamless deletion construction are improved. 60  References 1. Goffeau A, Barrell BG, Bussey H, Davis RW, Dujon B, Feldmann H, et al. Life with 6000 genes. Science. 1996;274:546, 63-7. 2. Hughes TR, Robinson MD, Mitsakakis N, Johnston M. The promise of functional genomics: completing the encyclopedia of a cell. Curr Opin Microbiol. 2004;7:546-54. 3. Brachmann CB, Davies A, Cost GJ, Caputo E, Li J, Hieter P, et al. Designer deletion strains derived from Saccharomyces cerevisiae S288C: a useful set of strains and plasmids for PCR-mediated gene disruption and other applications. Yeast. 1998;14:115-32. 4. Winzeler EA, Shoemaker DD, Astromoff A, Liang H, Anderson K, Andre B, et al. Functional characterization of the S. cerevisiae genome by gene deletion and parallel analysis. Science. 1999;285:901-6. 5. Giaever G, Chu AM, Ni L, Connelly C, Riles L, Veronneau S, et al. Functional profiling of the Saccharomyces cerevisiae genome. Nature. 2002;418:387-91. 6. Giaever G, Nislow C. The yeast deletion collection: a decade of functional genomics. Genetics. 2014;197:451-65. 7. Pierce SE, Fung EL, Jaramillo DF, Chu AM, Davis RW, Nislow C, et al. A unique and universal molecular barcode array. Nat Methods. 2006;3:601-3. 8. Pierce SE, Davis RW, Nislow C, Giaever G. Genome-wide analysis of barcoded Saccharomyces cerevisiae gene-deletion mutants in pooled cultures. Nat Protoc. 2007;2:2958-74. 9. Smith AM, Heisler LE, Mellor J, Kaper F, Thompson MJ, Chee M, et al. Quantitative phenotyping via deep barcode sequencing. Genome Res. 2009;19:1836-42. 61  10. Gresham D, Boer VM, Caudy A, Ziv N, Brandt NJ, Storey JD, et al. System-level analysis of genes and functions affecting survival during nutrient starvation in Saccharomyces cerevisiae. Genetics. 2011;187:299-317. 11. Robinson DG, Chen W, Storey JD, Gresham D. Design and analysis of Bar-seq experiments. G3 (Bethesda). 2014;4:11-8. 12. Hillenmeyer ME, Fung E, Wildenhain J, Pierce SE, Hoon S, Lee W, et al. The chemical genomic portrait of yeast: uncovering a phenotype for all genes. Science. 2008;320:362-5. 13. Costanzo M, Baryshnikova A, Bellay J, Kim Y, Spear ED, Sevier CS, et al. The genetic landscape of a cell. Science. 2010;327:425-31. 14. Lee AY, St Onge RP, Proctor MJ, Wallace IM, Nile AH, Spagnuolo PA, et al. Mapping the cellular response to small molecules using chemogenomic fitness signatures. Science. 2014;344:208-11. 15. Deutschbauer AM, Jaramillo DF, Proctor M, Kumm J, Hillenmeyer ME, Davis RW, et al. Mechanisms of haploinsufficiency revealed by genome-wide profiling in yeast. Genetics. 2005;169:1915-25. 16. Tong AH, Evangelista M, Parsons AB, Xu H, Bader GD, Page N, et al. Systematic genetic analysis with ordered arrays of yeast deletion mutants. Science. 2001;294:2364-8. 17. Ghaemmaghami S, Huh WK, Bower K, Howson RW, Belle A, Dephoure N, et al. Global analysis of protein expression in yeast. Nature. 2003;425:737-41. 18. Krogan NJ, Cagney G, Yu H, Zhong G, Guo X, Ignatchenko A, et al. Global landscape of protein complexes in the yeast Saccharomyces cerevisiae. Nature. 2006;440:637-43. 62  19. Schuldiner M, Collins SR, Thompson NJ, Denic V, Bhamidipati A, Punna T, et al. Exploration of the function and organization of the yeast early secretory pathway through an epistatic miniarray profile. Cell. 2005;123:507-19. 20. Boone C, Bussey H, Andrews BJ. Exploring genetic interactions and networks with yeast. Nat Rev Genet. 2007;8:437-49. 21. Cherry JM, Hong EL, Amundsen C, Balakrishnan R, Binkley G, Chan ET, et al. Saccharomyces Genome Database: the genomics resource of budding yeast. Nucleic Acids Res. 2012;40:D700-5. 22. Mortimer RK, Johnston JR. Genealogy of principal strains of the yeast genetic stock center. Genetics. 1986;113:35-43. 23. Corbacho I, Teixido F, Velazquez R, Hernandez LM, Olivero I. Standard YPD, even supplemented with extra nutrients, does not always compensate growth defects of Saccharomyces cerevisiae auxotrophic strains. Antonie Van Leeuwenhoek. 2011;99:591-600. 24. Cohen R, Engelberg D. Commonly used Saccharomyces cerevisiae strains (e.g. BY4741, W303) are growth sensitive on synthetic complete medium due to poor leucine uptake. FEMS Microbiol Lett. 2007;273:239-43. 25. Smith DL, Jr., Maharrey CH, Carey CR, White RA, Hartman JLt. Gene-nutrient interaction markedly influences yeast chronological lifespan. Exp Gerontol. 2016. 26. Fabrizio P, Hoon S, Shamalnasab M, Galbani A, Wei M, Giaever G, et al. Genome-wide screen in Saccharomyces cerevisiae identifies vacuolar protein sorting, autophagy, biosynthetic, and tRNA methylation genes involved in life span regulation. PLoS Genet. 2010;6:e1001024. 27. Matecic M, Smith DL, Pan X, Maqani N, Bekiranov S, Boeke JD, et al. A microarray-based genetic screen for yeast chronological aging factors. PLoS Genet. 2010;6:e1000921. 63  28. Bauer BE, Rossington D, Mollapour M, Mamnun Y, Kuchler K, Piper PW. Weak organic acid stress inhibits aromatic amino acid uptake by yeast, causing a strong influence of amino acid auxotrophies on the phenotypes of membrane transporter mutants. Eur J Biochem. 2003;270:3189-95. 29. Mulleder M, Capuano F, Pir P, Christen S, Sauer U, Oliver SG, et al. A prototrophic deletion mutant collection for yeast metabolomics and systems biology. Nat Biotechnol. 2012;30:1176-8. 30. Chopra R, Sharma VM, Ganesan K. Elevated growth of Saccharomyces cerevisiae ATH1 null mutants on glucose is an artifact of nonmatching auxotrophies of mutant and reference strains. Appl Environ Microbiol. 1999;65:2267-8. 31. Ding J, Bierma J, Smith MR, Poliner E, Wolfe C, Hadduck AN, et al. Acetic acid inhibits nutrient uptake in Saccharomyces cerevisiae: auxotrophy confounds the use of yeast deletion libraries for strain improvement. Appl Microbiol Biotechnol. 2013;97:7405-16. 32. Kirsch DR, Whitney RR. Pathogenicity of Candida albicans auxotrophic mutants in experimental infections. Infect Immun. 1991;59:3297-300. 33. Alam MT, Zelezniak A, Mulleder M, Shliaha P, Schwarz R, Capuano F, et al. The metabolic background is a global player in Saccharomyces gene expression epistasis. Nat Microbiol. 2016;1:15030. 34. Teng X, Dayhoff-Brannigan M, Cheng WC, Gilbert CE, Sing CN, Diny NL, et al. Genome-wide consequences of deleting any single gene. Mol Cell. 2013;52:485-94. 35. Mulleder M, Calvani E, Alam MT, Wang RK, Eckerstorfer F, Zelezniak A, et al. Functional Metabolomics Describes the Yeast Biosynthetic Regulome. Cell. 2016;167:553-65 e12. 64  36. Cooper SJ, Finney GL, Brown SL, Nelson SK, Hesselberth J, MacCoss MJ, et al. High-throughput profiling of amino acids in strains of the Saccharomyces cerevisiae deletion collection. Genome Res. 2010;20:1288-96. 37. Guo J, Liu H, Zheng J. SynLethDB: synthetic lethality database toward discovery of selective and sensitive anticancer drug targets. Nucleic Acids Res. 2016;44:D1011-7. 38. VanderSluis B, Hess DC, Pesyna C, Krumholz EW, Syed T, Szappanos B, et al. Broad metabolic sensitivity profiling of a prototrophic yeast deletion collection. Genome Biol. 2014;15:R64. 39. Gibney PA, Lu C, Caudy AA, Hess DC, Botstein D. Yeast metabolic and signaling genes are required for heat-shock survival and have little overlap with the heat-induced genes. Proc Natl Acad Sci U S A. 2013;110:E4393-402. 40. Kuepfer L, Sauer U, Blank LM. Metabolic functions of duplicate genes in Saccharomyces cerevisiae. Genome Res. 2005;15:1421-30. 41. Dudley AM, Janse DM, Tanay A, Shamir R, Church GM. A global view of pleiotropy and phenotypically derived gene function in yeast. Mol Syst Biol. 2005;1:2005 0001. 42. Lee W, St Onge RP, Proctor M, Flaherty P, Jordan MI, Arkin AP, et al. Genome-wide requirements for resistance to functionally distinct DNA-damaging agents. PLoS Genet. 2005;1:e24. 43. Parsons AB, Brost RL, Ding H, Li Z, Zhang C, Sheikh B, et al. Integration of chemical-genetic and genetic interaction data links bioactive compounds to cellular target pathways. Nat Biotechnol. 2004;22:62-9. 44. Smith AM, Durbic T, Oh J, Urbanus M, Proctor M, Heisler LE, et al. Competitive genomic screens of barcoded yeast libraries. J Vis Exp. 2011. 65  45. Yoshikawa K, Tanaka T, Ida Y, Furusawa C, Hirasawa T, Shimizu H. Comprehensive phenotypic analysis of single-gene deletion and overexpression strains of Saccharomyces cerevisiae. Yeast. 2011;28:349-61. 46. Huyer G, Kistler A, Nouvet FJ, George CM, Boyle ML, Michaelis S. Saccharomyces cerevisiae a-factor mutants reveal residues critical for processing, activity, and export. Eukaryot Cell. 2006;5:1560-70. 47. Chasse SA, Flanary P, Parnell SC, Hao N, Cha JY, Siderovski DP, et al. Genome-scale analysis reveals Sst2 as the principal regulator of mating pheromone signaling in the yeast Saccharomyces cerevisiae. Eukaryot Cell. 2006;5:330-46. 48. Yuen KW, Warren CD, Chen O, Kwok T, Hieter P, Spencer FA. Systematic genome instability screens in yeast and their potential relevance to cancer. Proc Natl Acad Sci U S A. 2007;104:3925-30. 49. Henry SA. Death resulting from fatty acid starvation in yeast. J Bacteriol. 1973;116:1293-303. 50. Boer VM, Amini S, Botstein D. Influence of genotype and nutrition on survival and metabolism of starving yeast. Proc Natl Acad Sci U S A. 2008;105:6930-5. 51. Unger MW, Hartwell LH. Control of cell division in Saccharomyces cerevisiae by methionyl-tRNA. Proc Natl Acad Sci U S A. 1976;73:1664-8. 52. Brauer MJ, Huttenhower C, Airoldi EM, Rosenstein R, Matese JC, Gresham D, et al. Coordination of growth rate, cell cycle, stress response, and metabolic activity in yeast. Mol Biol Cell. 2008;19:352-67. 53. Kohlhaw GB. Leucine biosynthesis in fungi: entering metabolism through the back door. Microbiol Mol Biol Rev. 2003;67:1-15, table of contents. 66  54. Hiltunen JK, Schonauer MS, Autio KJ, Mittelmeier TM, Kastaniotis AJ, Dieckmann CL. Mitochondrial fatty acid synthesis type II: more than just fatty acids. J Biol Chem. 2009;284:9011-5. 55. Ljungdahl PO, Daignan-Fornier B. Regulation of amino acid, nucleotide, and phosphate metabolism in Saccharomyces cerevisiae. Genetics. 2012;190:885-929. 56. Lacroute F, Pierard A, Grenson M, Wiame JM. The biosynthesis of carbamoyl phosphate in Saccharomyces cerevisiae. J Gen Microbiol. 1965;40:127-42. 57. Cheung-Ong K, Giaever G, Nislow C. DNA-damaging agents in cancer chemotherapy: serendipity and chemical biology. Chem Biol. 2013;20:648-59. 58. Hoepfner D, Helliwell SB, Sadlish H, Schuierer S, Filipuzzi I, Brachat S, et al. High-resolution chemical dissection of a model eukaryote reveals targets, pathways and gene functions. Microbiol Res. 2014;169:107-20. 59. Rebora K, Laloo B, Daignan-Fornier B. Revisiting purine-histidine cross-pathway regulation in Saccharomyces cerevisiae: a central role for a small molecule. Genetics. 2005;170:61-70. 60. Stearns T, Ma H, Botstein D. Manipulating yeast genome using plasmid vectors. Methods Enzymol. 1990;185:280-97. 61. Botstein D, Fink GR. Yeast: an experimental organism for modern biology. Science. 1988;240:1439-43. 62. Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B (Methodological). 1995;57:289-300. 67  63. Benaglia T, Chauveau D, Hunter D, R., Young D, S. mixtools: An R Package for Analyzing Finite Mixture Models. Journal of Statistical Software. 2009;32:1-29. 64. Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics. 2007;8:118-27. 65. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13:2498-504. 66. Bindea G, Mlecnik B, Hackl H, Charoentong P, Tosolini M, Kirilovsky A, et al. ClueGO: a Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics. 2009;25:1091-3. 67. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas. 1960;20:37-46. 68. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30:2114-20. 69. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25:1754-60. 70. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25:2078-9. 71. Quinlan AR, Hall IM. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 2010;26:841-2. 72. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26:139-40. 68  73. Lottaz C, Yang X, Scheid S, Spang R. OrderedList--a bioconductor package for detecting similarity in ordered gene lists. Bioinformatics. 2006;22:2315-6. 74. Baruffini E, Lodi T, Dallabona C, Foury F. A single nucleotide polymorphism in the DNA polymerase gamma gene of Saccharomyces cerevisiae laboratory strains is responsible for increased mitochondrial DNA mutability. Genetics. 2007;177:1227-31. 75. Dimitrov LN, Brem RB, Kruglyak L, Gottschling DE. Polymorphisms in multiple genes contribute to the spontaneous mitochondrial genome instability of Saccharomyces cerevisiae S288C strains. Genetics. 2009;183:365-83. 76. Gaisne M, Becam AM, Verdiere J, Herbert CJ. A 'natural' mutation in Saccharomyces cerevisiae strains derived from S288c affects the complex regulatory gene HAP1 (CYP1). Curr Genet. 1999;36:195-200. 77. Liu H, Styles CA, Fink GR. Saccharomyces cerevisiae S288C has a mutation in FLO8, a gene required for filamentous growth. Genetics. 1996;144:967-78. 78. Roop JI, Chang KC, Brem RB. Polygenic evolution of a sugar specialization trade-off in yeast. Nature. 2016;530:336-9. 79. Alam MT, Zelezniak A, Mülleder M, Shliaha P, Schwarz R, Capuano F, et al. The metabolic background is a global player in Saccharomyces gene expression epistasis. Nature Microbiology. 2016;1:15030. 80. Campbell K, Vowinckel J, Mulleder M, Malmsheimer S, Lawrence N, Calvani E, et al. Self-establishing communities enable cooperative metabolite exchange in a eukaryote. Elife. 2015;4. 69  81. Solis-Escalante D, Kuijpers NG, Bongaerts N, Bolat I, Bosman L, Pronk JT, et al. amdSYM, a new dominant recyclable marker cassette for Saccharomyces cerevisiae. FEMS Yeast Res. 2013;13:126-39. 82. Guldener U, Heck S, Fielder T, Beinhauer J, Hegemann JH. A new efficient gene disruption cassette for repeated use in budding yeast. Nucleic Acids Res. 1996;24:2519-24. 83. Gueldener U, Heinisch J, Koehler GJ, Voss D, Hegemann JH. A second set of loxP marker cassettes for Cre-mediated multiple gene knockouts in budding yeast. Nucleic Acids Res. 2002;30:e23. 84. Sauer B. Functional expression of the cre-lox site-specific recombination system in the yeast Saccharomyces cerevisiae. Mol Cell Biol. 1987;7:2087-96. 85. Carvalho A, Pereira F, Johansson B. The MX4blaster cassette: repeated and clean Saccharomyces cerevisiae genome modification using the genome-wide deletion collection. FEMS Yeast Res. 2013;13:711-9. 86. Mulleder M, Campbell K, Matsarskaia O, Eckerstorfer F, Ralser M. Saccharomyces cerevisiae single-copy plasmids for auxotrophy compensation, multiple marker selection, and for designing metabolically cooperating communities. F1000Res. 2016;5:2351. 87. Zhang GC, Kong, II, Kim H, Liu JJ, Cate JH, Jin YS. Construction of a quadruple auxotrophic mutant of an industrial polyploid saccharomyces cerevisiae strain by using RNA-guided Cas9 nuclease. Appl Environ Microbiol. 2014;80:7694-701. 88. Alexander WG, Doering DT, Hittinger CT. High-efficiency genome editing and allele replacement in prototrophic and wild strains of Saccharomyces. Genetics. 2014;198:859-66. 70  89. Deutschbauer AM, Williams RM, Chu AM, Davis RW. Parallel phenotypic analysis of sporulation and postgermination growth in Saccharomyces cerevisiae. Proc Natl Acad Sci U S A. 2002;99:15530-5. 90. Deutschbauer AM, Davis RW. Quantitative trait loci mapped to single-nucleotide resolution in yeast. Nat Genet. 2005;37:1333-40. 91. Contamine V, Picard M. Maintenance and integrity of the mitochondrial genome: a plethora of nuclear genes in the budding yeast. Microbiol Mol Biol Rev. 2000;64:281-315. 92. Merz S, Westermann B. Genome-wide deletion mutant analysis reveals genes required for respiratory growth, mitochondrial genome maintenance and mitochondrial protein synthesis in Saccharomyces cerevisiae. Genome Biol. 2009;10:R95. 93. Warringer J, Zorgo E, Cubillos FA, Zia A, Gjuvsland A, Simpson JT, et al. Trait variation in yeast is defined by population history. PLoS Genet. 2011;7:e1002111. 94. Bergstrom A, Simpson JT, Salinas F, Barre B, Parts L, Zia A, et al. A high-definition view of functional genetic variation from natural yeast genomes. Mol Biol Evol. 2014;31:872-88. 95. Ryan O, Shapiro RS, Kurat CF, Mayhew D, Baryshnikova A, Chin B, et al. Global gene deletion analysis exploring yeast filamentous growth. Science. 2012;337:1353-6. 96. Varela C, Kutyna DR, Solomon MR, Black CA, Borneman A, Henschke PA, et al. Evaluation of gene modification strategies for the development of low-alcohol-wine yeasts. Appl Environ Microbiol. 2012;78:6068-77. 97. Schwarzmuller T, Ma B, Hiller E, Istel F, Tscherner M, Brunke S, et al. Systematic phenotyping of a large-scale Candida glabrata deletion collection reveals novel antifungal tolerance genes. PLoS Pathog. 2014;10:e1004211.71  Appendices Appendix A  GO enrichments and gene lists from chapter 2 Table 3 Significantly enriched GO biological processes for 882 gene deletion strains missing from the BCprot collection (Bonferroni-corrected P-values) GO Biological Process # Genes % Associated Genes P-value Associated Genes Found amino acid activation [GO:0043038]  16 80 3.9E-07 FMT1, PET112, SLM5, MSW1, AIM10, GTF1, MSM1, DIA4, MSR1, MST1, NAM2, HER2, MSE1, ISM1, MSD1, MSF1 tRNA aminoacylation [GO:0043039] 16 80 3.9E-07 FMT1, PET112, SLM5, MSW1, AIM10, GTF1, MSM1, DIA4, MSR1, MST1, NAM2, HER2, MSE1, ISM1, MSD1, MSF1 cellular amino acid biosynthetic process [GO:0008652] 48 37 2.4E-05 MET8, HIS7, ARO4, HIS4, THR4, TRP1, LYS14, HOM2, TRP4, GLY1, HOM3, HIS1, TRP2, MET6, LPD1, HIS2, MET10, LEU1, TRP5, ARO2, LYS5, CYS4, ADE3, SER2, ARG4, THR1, PAN5, HIS6, LYS12, HIS5, LYS1, ARG2, ARG3, MDE1, CBF1, CPA2, HOM6, TRP3, MET1, AAT2, MET17, ACO1, ARG7, LYS9, ARG1, MET22, SER1, ARO7 tRNA aminoacylation for protein translation [GO:0006418] 13 76 3.6E-05 FMT1, SLM5, MSW1, AIM10, MSM1, DIA4, MSR1, MST1, NAM2, MSE1, ISM1, MSD1, MSF1 positive regulation of cellular protein metabolic process [GO:0032270] 23 51 5.8E-05 CBP6, RPS9B, PAF1, CBS1, STE7, STE5, CBS2, ATP22, ADA2, PET122, DOC1, RTF1, PET54, STE20, ASF1, ELM1, PET309, MSS51, BUR2, STE11, CTK3, PET111, PET494 mitochondrial RNA metabolic process [GO:0000959] 14 70 7.1E-05 SLM5, SLM3, MSW1, RPO41, CCM1, DIA4, MSR1, CBP1, MST1, DSS1, MSE1, ISM1, MSD1, MSF1 regulation of mitochondrial translation [GO:0070129] 12 75 1.6E-04 CBP6, CBS1, CBS2, ATP22, PET122, PET54, COA3, PET309, MSS51, COX14, PET111, PET494 RNA processing [GO:0006396] 66 30 3.3E-04 CCR4, FMT1, RPS8A, TRM7, CBP6, RPS9B, PAF1, PAT1, PTC1, SLM3, MTF2, SIT4, DHH1, RPS11A, HPR1, MSS116, REF2, LSM6, DST1, NCS6, RTF1, SLX9, CCM1, NSR1, ELP2, RPS0A, PET54, ZUO1, CBP2, SLT2, SSZ1, LRP1, IKI1, URM1, NOT3, RPS24B, IST3, MSL1, MRS1, LSM1, RPS21B, CBP1, REX2, MSS51, MRPL15, NAM2, TRM9, RPS1B, CTK3, STO1, CUS2, POP2, PAP2, TSR3, BUD21, RPS7A, MRM1, GEP3, MOD5, MFM1, ELP3, CBC2, LEA1, THP3, MSF1, SNT309 72  mRNA processing [GO:0006397] 32 38 1.2E-03 CCR4, CBP6, PAF1, PAT1, MTF2, DHH1, HPR1, MSS116, REF2, LSM6, DST1, RTF1, CCM1, PET54, CBP2, SLT2, NOT3, IST3, MSL1, MRS1, LSM1, CBP1, MSS51, CTK3, STO1, CUS2, POP2, MFM1, CBC2, LEA1, THP3, SNT309 tRNA aminoacylation for mitochondrial protein translation [GO:0070127] 9 82 1.3E-03 SLM5, MSW1, DIA4, MSR1, MST1, MSE1, ISM1, MSD1, MSF1 ubiquinone biosynthetic process [GO:0006744] 8 89 1.4E-03 COQ1, COQ4, COQ6, COQ9, COQ5, COQ2, COQ3, CAT5 respiratory chain complex IV assembly [GO:0008535] 12 63 2.5E-03 SCO1, PET100, COX20, PET117, COA1, COX16, COA3, PET191, COX17, COX19, COX14, RCF2 ncRNA metabolic process [GO:0034660] 52 31 2.5E-03 FMT1, RPS8A, PET112, TRM7, RPS9B, PAF1, SLM5, PTC1, SLM3, SIT4, RPS11A, REF2, MSW1, LSM6, AIM10, NCS6, RTF1, SLX9, GTF1, CCM1, NSR1, MSM1, ELP2, RPS0A, ZUO1, DIA4, SSZ1, LRP1, MSR1, IKI1, URM1, RPS24B, RTT101, RPS21B, MST1, REX2, NAM2, TRM9, RPS1B, HER2, MSE1, PAP2, TSR3, BUD21, RPS7A, MRM1, GEP3, MOD5, ISM1, ELP3, MSD1, MSF1 aspartate family amino acid metabolic process [GO:0009066] 23 43 2.9E-03 MET8, CHA1, THR4, LYS14, HOM2, GLY1, HOM3, MET6, MET10, LYS5, ADE3, THR1, LYS12, MET18, LYS1, MDE1, CBF1, HOM6, MET1, AAT2, MET17, LYS9, MET22 aspartate family amino acid biosynthetic process [GO:0009067] 20 45 3.3E-03 MET8, THR4, LYS14, HOM2, HOM3, MET6, MET10, LYS5, ADE3, THR1, LYS12, LYS1, MDE1, CBF1, HOM6, MET1, AAT2, MET17, LYS9, MET22 positive regulation of mitochondrion organization [GO:0010822] 9 75 4.1E-03 CBP6, CBS1, CBS2, ATP22, PET122, PET54, MSS51, PET111, PET494 positive regulation of mitochondrial translation [GO:0070131] 9 75 4.1E-03 CBP6, CBS1, CBS2, ATP22, PET122, PET54, MSS51, PET111, PET494 mitochondrial respiratory chain complex IV assembly [GO:0033617] 11 65 4.1E-03 PET100, COX20, PET117, COA1, COX16, COA3, PET191, COX17, COX19, COX14, RCF2 mitochondrial respiratory chain complex assembly [GO:0033108] 15 52 5.6E-03 CBP6, MRP10, PET100, COX20, BCS1, QCR7, PET117, COA1, COX16, COA3, PET191, COX17, COX19, COX14, RCF2 73  aromatic amino acid family biosynthetic process [GO:0009073] 8 73 1.7E-02 ARO4, TRP1, TRP4, TRP2, TRP5, ARO2, TRP3, ARO7 chorismate metabolic process [GO:0046417] 8 73 1.7E-02 ARO4, TRP1, TRP4, TRP2, TRP5, ARO2, TRP3, ARO7 regulation of translation [GO:0006417] 28 35 2.5E-02 TPD3, FUN12, FES1, CBP6, RPS9B, PAT1, CBS1, DHH1, MTQ2, MSS116, CBS2, ATP22, SNF1, PET122, GCN1, PET54, ZUO1, SSZ1, COA3, PET309, MSS51, CTK3, COX14, AEP1, ASC1, PET111, AEP2, PET494 threonine metabolic process [GO:0006566] 7 78 2.5E-02 CHA1, THR4, HOM2, GLY1, HOM3, THR1, HOM6 pheromone-dependent signal transduction involved in conjugation with cellular fusion [GO:0000750] 14 48 2.7E-02 FUS3, STE50, FYV5, PTC1, STE7, STE5, AKR1, MFA1, STE2, STE20, GPA1, SCP160, STE11, STE4 positive regulation of translation [GO:0045727] 12 52 3.1E-02 CBP6, RPS9B, CBS1, CBS2, ATP22, PET122, PET54, PET309, MSS51, CTK3, PET111, PET494 tryptophan biosynthetic process [GO:0000162] 5 100 3.2E-02 TRP1, TRP4, TRP2, TRP5, TRP3 threonine biosynthetic process [GO:0009088] 5 100 3.2E-02 THR4, HOM2, HOM3, THR1, HOM6 indolalkylamine biosynthetic process [GO:0046219] 5 100 3.2E-02 TRP1, TRP4, TRP2, TRP5, TRP3 intralumenal vesicle formation [GO:0070676] 5 100 3.2E-02 DID4, VPS24, SNF7, VPS20, VPS4 tRNA metabolic process [GO:0006399] 29 34 4.0E-02 FMT1, PET112, TRM7, SLM5, PTC1, SLM3, SIT4, MSW1, LSM6, AIM10, NCS6, GTF1, MSM1, ELP2, DIA4, MSR1, IKI1, URM1, MST1, NAM2, TRM9, HER2, MSE1, PAP2, MOD5, ISM1, ELP3, MSD1, MSF1    74  Table 4 Gene deletion strains present in the HLU signature ORF Gene SGD Description[21] YOR128C ADE2 Phosphoribosylaminoimidazole carboxylase YDR530C APA2 Diadenosine 5',5'''-P1,P4-tetraphosphate phosphorylase II; YHR018C ARG4 Argininosuccinate lyase YGL148W ARO2 Bifunctional chorismate synthase and flavin reductase; protein abundance increases in response to DNA replication stress YDL088C ASM4 FG-nucleoporin component of central core of nuclear pore complex (NPC) YPR049C ATG11 Adapter protein for pexophagy and the Cvt targeting pathway; directs receptor-bound cargo to the phagophore assembly site (PAS) for packaging into vesicles YPL166W ATG29 Autophagy-specific protein; required for recruiting other ATG proteins to the pre-autophagosomal structure (PAS); relocalizes from nucleus to cytoplasmic foci upon DNA replication stress YKL016C ATP7 Subunit d of the stator stalk of mitochondrial F1F0 ATP synthase YOR299W BUD7 Member of the ChAPs family (Chs5p-Arf1p-binding proteins); ChAPs family proteins form the exomer complex with Chs5p to mediate export of specific cargo proteins from the Golgi to the plasma membrane;  YMR198W CIK1 Kinesin-associated protein; required for both karyogamy and mitotic spindle organization YPR119W CLB2 B-type cyclin involved in cell cycle progression; activates Cdc28p to promote the transition from G2 to M phase YPR120C CLB5 B-type cyclin involved in DNA replication during S phase; activates Cdc28p to promote initiation of DNA synthesis; functions in formation of mitotic spindles  YGR167W CLC1 Clathrin light chain; subunit of the major coat protein involved in intracellular protein transport and endocytosis YMR048W CSM3 Replication fork associated factor; required for stable replication fork pausing; component of the DNA replication checkpoint pathway; required for accurate chromosome segregation during meiosis; forms nuclear foci upon DNA replication stress YPL018W CTF19 Outer kinetochore protein, needed for accurate chromosome segregation; component of the kinetochore sub-complex COMA (Ctf19p, Okp1p, Mcm21p, Ame1p) that functions as a platform for kinetochore assembly; required for the 75  spindle assembly checkpoint YPR135W CTF4 Chromatin-associated protein; required for sister chromatid cohesion; interacts with DNA polymerase alpha (Pol1p) and may link DNA synthesis to sister chromatid cohesion YAL012W CYS3 Cystathionine gamma-lyase; catalyzes one of the two reactions involved in the transsulfuration pathway that yields cysteine from homocysteine with the intermediary formation of cystathionine; protein abundance increases in response to DNA replication stress YGR155W CYS4 Cystathionine beta-synthase; catalyzes synthesis of cystathionine from serine and homocysteine YKL204W EAP1 eIF4E-associated protein, competes with eIF4G for binding to eIF4E; accelerates mRNA degradation by promoting decapping, inhibits cap-dependent translation; functions independently of eIF4E to maintain genetic stability; plays a role in cell growth YOR144C ELG1 Subunit of an alternative replication factor C complex; important for DNA replication and genome integrity; suppresses spontaneous DNA damage; involved in homologous recombination-mediated repair and telomere homeostasis YHR059W FYV4 Protein of unknown function; required for survival upon exposure to K1 killer toxin YDL100C GET3 Guanine nucleotide exchange factor for Gpa1p; amplifies G protein signaling; functions as a chaperone under ATP-depleted oxidative stress conditions; involved in ATP dependent Golgi to ER trafficking and insertion of tail-anchored (TA) proteins into ER membrane under non-stress conditions; protein abundance increases under DNA replication stress YMR032W HOF1 Protein that regulates actin cytoskeleton organization; required for cytokinesis, actin cable organization, and secretory vesicle trafficking; localized to bud neck; regulates actomyosin ring dynamics and septin localization YIL110W HPM1 AdoMet-dependent methyltransferase YDR138W HPR1 Subunit of THO/TREX complexes; this complex couple transcription elongation with mitotic recombination and with mRNA metabolism and export, subunit of an RNA Pol II complex; regulates lifespan; involved in telomere maintenance YJR122W IBA57 Protein involved in incorporating iron-sulfur clusters into proteins YBR107C IML3 Outer kinetochore protein and component of the Ctf19 complex; involved in the establishment of pericentromeric cohesion during mitosis; prevents non-disjunction of sister chromatids during meiosis II YDR123C INO2 Transcription factor; component of the heteromeric Ino2p/Ino4p basic helix-loop-helix transcription activator that binds inositol/choline-responsive elements (ICREs) 76  YOR135C IRC14 Dubious open reading frame; unlikely to encode a functional protein; partially overlaps YOR136W; null mutant displays increased levels of spontaneous Rad52 foci YPL017C IRC15 Microtubule associated protein; regulates microtubule dynamics; required for accurate meiotic chromosome segregation YLL027W ISA1 Protein required for maturation of mitochondrial [4Fe-4S] proteins; functions in a complex with Isa2p and possibly Iba57p; isa1 deletion causes loss of mitochondrial DNA and respiratory deficiency YPR067W ISA2 Protein required for maturation of mitochondrial [4Fe-4S] proteins; functions in a complex with Isa1p and possibly Iba57p YDR475C JIP4 Protein of unknown function; previously annotated as two separate ORFs, YDR474C and YDR475C, which were merged YER110C KAP123 Karyopherin beta; mediates nuclear import of ribosomal proteins prior to assembly into ribosomes and import of histones H3 and H4 YPR141C KAR3 Minus-end-directed microtubule motor; functions in mitosis and meiosis, localizes to the spindle pole body and localization is dependent on functional Cik1p YHL003C LAG1 Ceramide synthase component; involved in synthesis of ceramide from C26(acyl)-coenzyme A and dihydrosphingosine or phytosphingosine; forms ER foci upon DNA replication stress YBL006C LDB7 Component of the RSC chromatin remodeling complex; interacts with Rsc3p, Rsc30p, Npl6p, and Htl1p to form a module for a broad range of RSC functions YNL147W LSM7 Lsm (Like Sm) protein; part of heteroheptameric complexes (Lsm2p-7p and either Lsm1p or 8p): cytoplasmic Lsm1p complex involved in mRNA decay; protein abundance increases and forms cytoplasmic foci in response to DNA replication stress YAL024C LTE1 Protein similar to GDP/GTP exchange factors; without detectable GEF activity YDR318W MCM21 Component of the kinetochore sub-complex COMA; COMA (Ctf19p, Okp1p, Mcm21p, Ame1p) bridges kinetochore subunits in contact with centromeric DNA with subunits bound to microtubules during kinetochore assembly; involved in minichromosome maintenance YCL061C MRC1 S-phase checkpoint protein required for DNA replication; couples DNA helicase and polymerase; interacts with and stabilizes Pol2p at stalled replication forks during stress YKL057C NUP120 Subunit of the Nup84p subcomplex of the nuclear pore complex (NPC); contributes to nucleocytoplasmic transport and NPC biogenesis and is involved in establishment of a normal nucleocytoplasmic concentration gradient of the GTPase Gsp1p; also plays roles in several processes that may require localization of genes or chromosomes at the nuclear periphery, including double-strand break repair, transcription and chromatin silencing YBL079W NUP170 Subunit of the inner ring of the nuclear pore complex (NPC); contributes to NPC assembly and nucleocytoplasmic transport 77  YML103C NUP188 Subunit of the inner ring of the nuclear pore complex (NPC); contributes to NPC organization and nucleocytoplasmic transport YDL089W NUR1 Protein involved in regulation of mitotic exit; dephosphorylation target of Cdc14p in anaphase, which promotes timely rDNA segregation and allows mitotic progression; null mutant causes increase in unequal sister-chromatid exchange YDR488C PAC11 Dynein intermediate chain, microtubule motor protein; required for intracellular transport and cell division YDR479C PEX29 Peroxisomal integral membrane peroxin; involved in the regulation of peroxisomal size, number and distribution; forms ER foci upon DNA replication stress YPL144W POC4 Component of a heterodimeric Poc4p-Irc25p chaperone; involved in assembly of alpha subunits into the 20S proteasome; may regulate formation of proteasome isoforms with alternative subunits under different conditions YPL079W RPL21B Ribosomal 60S subunit protein L21B YDL081C RPP1A Ribosomal stalk protein P1 alpha; involved in the interaction between translational elongation factors and the ribosome YGR056W RSC1 Component of the RSC chromatin remodeling complex YLR357W RSC2 Component of the RSC chromatin remodeling complex YDR494W RSM28 Mitochondrial ribosomal protein of the small subunit; genetic interactions suggest a possible role in promoting translation initiation YMR060C SAM37 Component of the Sorting and Assembly Machinery (SAM) complex YGL066W SGF73 Subunit of DUBm module of SAGA and SLIK YML058W SML1 Ribonucleotide reductase inhibitor; involved in regulating dNTP production; regulated by Mec1p and Rad53p during DNA damage and S phase YPL002C SNF8 Component of the ESCRT-II complex; ESCRT-II is involved in ubiquitin-dependent sorting of proteins into the endosome YER161C SPT2 Protein involved in negative regulation of transcription; required for RNA polyadenylation; exhibits regulated interactions with both histones and SWI-SNF components YBR081C SPT7 Subunit of the SAGA transcriptional regulatory complex; involved in proper assembly of the complex; also present as a C-terminally truncated form in the SLIK/SALSA transcriptional regulatory complex YLR055C SPT8 Subunit of the SAGA transcriptional regulatory complex 78  YHR041C SRB2 Subunit of the RNA polymerase II mediator complex; associates with core polymerase subunits to form the RNA polymerase II holoenzyme; general transcription factor involved in telomere maintenance YDR443C SSN2 Subunit of the RNA polymerase II mediator complex; associates with core polymerase subunits to form the RNA polymerase II holoenzyme; essential for transcriptional regulation YNL273W TOF1 Subunit of a replication-pausing checkpoint complex; Tof1p-Mrc1p-Csm3p acts at the stalled replication fork to promote sister chromatid cohesion after DNA damage, facilitating gap repair of damaged DNA; interacts with the MCM helicase YLR420W URA4 Dihydroorotase YNL107W YAF9 Subunit of NuA4 histone H4 acetyltransferase and SWR1 complexes; may function to antagonize silencing near telomeres YAR029W YAR029W Member of DUP240 gene family but contains no transmembrane domains YDR149C YDR149C Dubious open reading frame; unlikely to encode a functional protein; overlaps the verified gene NUM1/YDR150W YDR491C YDR491C Dubious open reading frame; unlikely to encode a functional protein YDR509W YDR509W Dubious open reading frame; unlikely to encode a functional protein YGL109W YGL109W Dubious open reading frame; unlikely to encode a functional protein; overlaps the uncharacterized gene YGL108C YGL117W YGL117W Putative protein of unknown function YDR451C YHP1 Homeobox transcriptional repressor; binds Mcm1p and early cell cycle box (ECB) elements of cell cycle regulated genes, thereby restricting ECB-mediated transcription to the M/G1 interval YPL182C YPL182C Dubious open reading frame; unlikely to encode a functional protein; partially overlaps the verified gene CTI6/YPL181W    79  Table 5 Significantly enriched GO biological processes for 73 gene deletion strains in the HLU signature (Bonferroni-corrected P-values) GO Biological Process # Genes % Associated Genes P-value Associated Genes Found M phase of mitotic cell cycle [GO:0000087] 17 15 5.80E-11 LTE1, NUP170, IML3, MRC1, CDC10, ASM4, SAC3, MCM21, DYN1, CSM3, CIK1, TOF1, ELG1, IRC15, CTF19, CLB2, KAR3 mitosis [GO:0007067] 16 16 1.30E-10 LTE1, NUP170, IML3, MRC1, ASM4, SAC3, MCM21, DYN1, CSM3, CIK1, TOF1, ELG1, IRC15, CTF19, CLB2, KAR3 nuclear division [GO:0000280] 16 16 1.50E-10 LTE1, NUP170, IML3, MRC1, ASM4, SAC3, MCM21, DYN1, CSM3, CIK1, TOF1, ELG1, IRC15, CTF19, CLB2, KAR3 mitotic sister chromatid segregation [GO:0000070] 11 26 1.40E-09 IML3, MRC1, MCM21, DYN1, CSM3, CIK1, TOF1, ELG1, IRC15, CTF19, KAR3 sister chromatid segregation [GO:0000819] 11 24 3.10E-09 IML3, MRC1, MCM21, DYN1, CSM3, CIK1, TOF1, ELG1, IRC15, CTF19, KAR3 mitotic sister chromatid cohesion [GO:0007064] 8 36 4.40E-08 IML3, MRC1, MCM21, CSM3, TOF1, ELG1, CTF19, KAR3 sister chromatid cohesion [GO:0007062] 8 29 3.80E-07 IML3, MRC1, MCM21, CSM3, TOF1, ELG1, CTF19, KAR3 mRNA transport [GO:0051028] 8 14 1.20E-04 NUP170, ASM4, HPR1, SAC3, KAP123, SGF73, NUP120, NUP188 regulation of G2/M transition of mitotic cell cycle [GO:0010389] 5 31 1.80E-04 MRC1, CSM3, TOF1, CLB2, CLB5 microtubule-based transport [GO:0010970] 5 31 1.80E-04 PAC11, DYN1, CIK1, DYN3, KAR3 nuclear migration along microtubule [GO:0030473] 5 31 1.80E-04 PAC11, DYN1, CIK1, DYN3, KAR3 organelle transport along microtubule [GO:0072384] 5 31 1.80E-04 PAC11, DYN1, CIK1, DYN3, KAR3   80  Appendix B  Access to data Online: The datasets generated for this thesis and related research article will be available in the BioProject repository PRJNA338880 upon publication. The microarray data is deposited at GSE89761 and will be available upon publication.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0340607/manifest

Comment

Related Items