IGES 2009: Overcoming Data Quality and Copy Number Detection Issues in Genome-Wide CNV Association Studies

Presenter: Dr. Christophe Lambert, Golden helix CEO and Co-Chair of the FDA MAQC CNV Team

Date: October 19, 2009

Duration: 20 Minutes

ACCESS THIS WEBCAST:

Abstract

Incorporating copy number variations into genome-wide association studies promises to explain more of the heritability of common diseases than that accounted for by SNPs alone. This potential goldmine however, has been plagued by myriad of technical and experimental challenges. We examine the most persistent issues observed in over 20 CNV GWAS studies conducted by us and our collaborators. These include massive amounts of raw data, batch effects, genomic waves, mosaicism, T-cell artifacts, and poor signal-to-noise ratios, all of which can lead to false positive and negative CNV detection and subsequent association findings.

To address these issues we describe a novel principal component analysis approach that simultaneously corrects for batch and wave effects and population stratification, while significantly improving signal-to-noise ratios. We then describe optimal segmentation methods which use dynamic programming to detect copy number segment boundaries on either a per-sample (univariate) or a multi-sample (multivariate) basis. Unlike Hidden Markov Model methods, which assume the means of different copy number states are consistent, optimal segmenting methods properly delineate segment boundaries in the presence of mosaicism, even at a single probe level, and with superior sensitivity and false discovery rates. We then outline several approaches to genome-wide scans for CNV association, demonstrating the utility of these methods on a series of large-scale GWAS.

About the Presenter

Dr. Christophe Lambert is the President and CEO of Golden Helix, Inc., a bioinformatics company he founded in Bozeman, MT in 1998. Dr. Lambert graduated with his Bachelors in Computer Science from Montana State University in 1992 and received his Ph.D. in Computer Science from Duke University in 1997. Dr. Lambert is also currently the co-chair of the Food and Drug Administration’s Genome Wide Copy Number Variation Data Analysis Team of the Microarray Quality Control Consortium.