On region based inference in genome wide association study

Sergey V. Malov¹, Alexey Antonik², Andrey K. Shevchenko³
¹St.-Petersburg State University, malovs@sm14820.spb.edu
²St.-Petersburg State University, alexey.antonik@gmail.com
³St.-Petersburg State University, andrey.k.shevchenko@gmail.com

We consider an advanced framework for genome wide association study (GWAS) based on the signal localization. Instead of looking for single genetic markers associated with the phenotype we attempt to discover regions of genetic markers, which are associated with phenotype. We focus on the localization by moving sums of negative logarithms of p-values, which is efficient for discovery of wide regions of genetic markers associated with phenotype. Particularly, we expect the method should be efficient for searching genes containing a number of genetic markers, if any reconstruction in a gene related to a phenotype. The method is implemented in GWATCH software for visualization and interpretation results of multiple statistical tests for genome associations. We discuss some features of the GWATCH software and apply them for HIV/AIDS cohort study from Botswana.

0 0 vote

Article Rating

12 Comments

Inline Feedbacks

View all comments

Alexander Zlobin

3 years ago

Thank you for the interesting topic! Is it right that your approach found a region with a lot of loci in high linkage disequilibrium? In standard prioritization protocol for GWAS always discover not only top SNP, but also SNPs with high LD with top SNP. And your approach looks like found a region with SNPs in high LD.