Open Collections

Khan, Almas

University of British Columbia

The Placentomics project aims to create a compendium of various ‘omics tools applied to the study of the human placenta. One of these ‘omics tools is methylomics data collected through the Illumina 850K (EPIC) array. However, there is lack of standardization of the statistical processing of the array data. In this thesis, I studied the technical, clinical, and biological factors that can influence DNAme differences and used these factors to create a standardized processing pipeline. I used 204 samples from three cohorts of pregnant women: 36 samples from a normative cohort (V-NORM), 64 from a SSRI exposed cohort (V-SSRI) and 105 samples from an acute stressor exposed cohort (QF2011). In chapter 2, I evaluated factors that contribute to DNA methylation variability and identified clinical gestational age, PlaNET derived cell type proportions and PlaNET derived genetic ancestry probabilities as important to include in the pipeline, in addition to other R DNAme analysis packages such as minfi and wateRmelon. I then used this pipeline to assess association between DNAme and two biological variables of interest: birth weight (sd) and placental efficiency (placental weight: fetal weight ratio). Three differentially methylated probes met significance (FDR<0.05) and strength cut-offs (delta beta >0.05 or delta beta <-0.05) in the placental efficiency model only . I further assessed these variables of interest in models that excluded cell type proportions or considered sex stratification. I found no probes that met both cut-offs in these models. In Chapter 3, I developed and standardized a set of metrics to select a normalization method for DNAme arrays. I evaluated 4 within-sample, 2 between-sample, and 1 method that had elements of both (between and within-sample) methods using these metrics. I also evaluated 3 combinations of methods (i.e., Noob and another method). I found a combination of between and within-sample methods (Dasen+Noob) to perform the best on these metrics in this data set. I also found that including Noob prior to another normalization method produced consistently good results on my metrics. Thus, through these studies, I provided a set of standardized tools for the analysis of DNAme data in the placenta.

Text

2022-07-12

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/82117

Bioinformatics

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/