Learning a Prior on Regulatory Potential from eQTL Data

Since genome-wide RNA expression data provide a detailed view of an organism's biological state, expression variation between genetically diverse individuals (eQTL) may provide important insights into the genetics of complex traits.  However with data from a relatively small number of individuals, it is difficult to distinguish true causal polymorphisms from the large number of possibilities.  The problem is particularly challenging in systems with significant linkage disequilibrium, where traits are often linked to large chromosomal regions containing many genes.  Here we present a novel method, Lirnet, that automatically learns a regulatory potential for each sequence variation, estimating how likely it is to have a significant effect on gene expression.  This regulatory potential is defined in terms of "regulatory features" – including gene function and the conservation, type and position of genetic polymorphisms – that are available for any organism.  The extent to which the different features influence the regulatory potential is learned automatically, making Lirnet readily applicable to different data sets, organisms, and feature sets. 

Learned Model

List of candidate expression regulators    DOWNLOAD

Supplementary Table S5    DOWNLOAD

Module genes DOWNLOAD

Regulatory programs    DOWNLOAD

Positions of genetic markers    DOWNLOAD

Meta-features

Lirnet Software

Our matlab code for Lirnet can be downloaded here.

Visualization Tool

Our visualization tool GenViewer can help the analysis on the resulting network.  The executable for Windows can be downloaded here.

Any question can be addressed to Su-In Lee (5219silee@cs.stanford.edu3391: remove #s)