chapter  6
28 Pages

The Graph-Guided Group Lasso for Genome-Wide Association Studies

WithZi Wang, Giovanni Montana

Conventional GWAs involve a case-control study design, in which the subjects are partitioned into case and control groups, and hypothesis tests (e.g., based on χ2 statistic) are carried out for each SNP to examine if the frequencies of the genotypes (commonly denoted as aa, Aa, and AA) are significantly altered between the case and control groups [5, 6]. Some limitations of this approach include: substantial estimation biases and spurious associations due to violation of the assumptions in case-control design [28]; insufficient sample size in comparison to the number of variables, typically sample sizes vary between a few hundred and a few thousand, whereas the number of SNPs to be investigated is about 1 million and may go beyond 10 million [6, 24]; and multiple testing issues [28].