FastLORS: Joint Modeling for eQTL Mapping in R

TL;DR

FastLORS is an efficient algorithm that improves the computational speed of joint modeling for eQTL mapping, making large-scale genetic data analysis more feasible without sacrificing accuracy.

Contribution

The paper introduces FastLORS, a proximal gradient-based method that significantly reduces computation time for eQTL mapping compared to the original LORS.

Findings

FastLORS achieves comparable results to LORS on HapMap data.

FastLORS reduces computational time substantially.

The method maintains accuracy while improving efficiency.

Abstract

Yang et al. (2013) introduced LORS, a method that jointly models the expression of genes, SNPs, and hidden factors for eQTL mapping. LORS solves a convex optimization problem and has guaranteed convergence. However, it can be computationally expensive for large datasets. In this paper we introduce Fast-LORS which uses the proximal gradient method to solve the LORS problem with significantly reduced computational burden. We apply Fast-LORS and LORS to data from the third phase of the International HapMap Project and obtain comparable results. Nevertheless, Fast-LORS shows substantial computational improvement compared to LORS.