A genetical genomics approach to genome scans for complex traits
MetadataShow full item record
In the past few years, genetical genomics - the combination of gene expression and genetic marker data, have been used to conduct genome-wide survey of the genetics of gene expression. However, most of studies depend on prior knowledge of the location of QTLs for a clinical trait. We have developed a new EM-algorithm based method for integrating expression information into genome scans for complex trait. That is, we desire to identify genetic loci underlying clinical traits when there is not any prior knowledge of trait loci. The goal of this method is test whether the controlling factor for a group of highly correlated genes is also a causative locus for a disease or other clinical trait. We use this as part of a new strategy for genome scans for complex traits. We first screen the data to identify the set of expression levels which are the most promising targets for association with the trait. We next use hierarchical clustering to identify groups of highly correlated transcripts. We apply our EM algorithm method to each gene group to test whether any controlling loci are also causative for the clinical trait. Expression QTLs for such gene groups become candidate loci for the clinical trait. Using both simulated and real data, we show this strategy to have excellent power relative to existing methods.