Constrained Functional Linear Model for Multi-Loci Genetic Mapping

• Activity Number: 359
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Jiayu Huang* and Song Wu
• Companies: SUNY Stony Brook and SUNY Stony Brook
• Keywords: Functional Linear Model ; B-splines ; constraints ; GWAS

Abstract

Most contemporary methods in GWAS only consider one genetic variant at a time, and very few of them try to address how the linkage disequilibrium among dense-typed linked genetic variants may impact the association power. Here we propose a novel functional linear model (FLM) based approach to perform a simultaneous association test of a block of highly linked SNPs with a phenotypic trait. The new model applies a set of constraints on the FLM to ensure its identifiability, and is therefore called the constrained FLM (cFLM). The cFLM is flexible in addressing the allelic correlation among grouped SNPs, as well as guaranteeing that the fitted functions are interpretable. The method is implemented with the B-splines, and parameter estimation employs the Augmented Lagrangian Algorithm. A hypothesis test is formulated to assess the genetic association by defining a test statistic following a mixture of chi-square distributions. Our simulation results show that the new method is effective in identifying disease-causal SNPs compared to three competing methods. We apply the proposed method to analyze the COGEND (The Collaborative Genetic Study of Nicotine Dependence) data.