Open Collections

Model-based clustering for aCGH data using variational EM

University of British Columbia

DNA copy number alterations (CNAs) are genetic changes that can produce adverse effects in numerous human diseases, including cancer. Copy number variations (of which CNAs are a subset) are a common phenomenon and not much is known about the nature of many of the mutations. By clustering patients according to CNA patterns, we can identify recurrent CNAs and understand molecular heterogeneity. This differs from normal distance-based clustering that doesn’t exploit the sequential structure of the data. Our approach is based on the hmmmix model introduced by [Sha08]. We show how it can be trained with variational methods to achieve better results and make it more flexible. We show how this allows for soft patient clusterings and how it partly addresses the difficult issue of determining the number of clusters to use. We compare the performance of our method with that of [Sha08] using their original benchmark test as well as with synthetic data generated from the hmmmix model itself. We show how our method can be parallelized and adapted to huge datasets.

Thesis/Dissertation

application/pdf

2009-08-11

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/11992

Computer Science

University of British Columbia

UBCV

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