Efficient MVN approach to identify eGenes in eQTL studies

eGene-MVN is an efficient and accurate multivariate normal sampling (MVN) approach to perform the multiple testing correction in eQTL studies. As the above figure shows, we first find a SNP with the minimum p-value among all SNPs in cis with a gene. We then perform multiple testing correction using our MVN approach to correct this p-value. This correct p-value is "eGene" p-value for this gene. Our approach is more than 100 times faster than the permutation test when sample size is 2,000. When sample size is small such as 70, our method is still at least 25 times faster than the permutation test. Our algorithm uses two scaling algorithms to make MVN approach have similar eGene p-values as those from the permutation test. For more information, please refer to the following manuscript we published:

Jae Hoon Sul, Towfique Raj, Simone de Jong, Paul I. W. de Bakker, Soumya Raychaudhuri, Roel A. Ophoff, Barbara E. Stranger, Eleazar Eskin, Buhm Han. “Accurate and fast multiple testing correction in eQTL studies.” The American Journal of Human Genetics, 2015. In Press.


May 22, 2015: Version 0.1 is posted.