Modeling Discovery of Functional SNPS from Genome Scale Data

• Activity Number: 25
• Type: Contributed
• Date/Time: Sunday, July 31, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Stephanie Hicks*+ and Sharon Plon and Marek Kimmel
• Companies: Rice University and Baylor College of Medicine and Rice University
• Keywords: missense mutation ; whole-exome sequencing ; SIFT ; PolyPhen-2 ; Align-GVGD ; Xvar

Abstract

Computational algorithms have been developed to predict the impact of missense mutations on protein structure and function. We compare predictions of the PolyPhen-2 and Xvar algorithms using native alignments on a set of missense mutations (n = 2416) extracted from whole-exome sequencing data of one individual which results in classification of each variant as either deleterious or neutral. When considering the predicted deleterious mutations PolyPhen-2 and Xvar agree on 33% compared to previously reported lower overlap of 17% between PolyPhen and SIFT [Chun and Fay 2009 Genome Res 19: 1553]. Interestingly, SIFT, Align-GVGD, PolyPhen-2 and Xvar agree on 68% using a smaller list of well characterized mutations known to be deleterious or neutral [Hicks et al. 2011 Human Mut to appear]. Reconciling these outcomes is important for the practice of functionality prediction. The varying degree of overlap is likely to be caused by the nonuniform density and differences in functional importance of different SNPs. We develop a parametric model for distributions of SNP density and functionality to evaluate the varying level of agreements between studies when employing different algorithms.