Each set of gene expression correlates with P-value < 0.001 for gene–phenotype association was subject to combinatorial analysis to identify those genes that were directly correlated to multiple phenotypes. Positional candidates defined as those which reside within an one-LOD confidence interval of significant (P-value < 0.05) or suggestive (P-value < 0.63) QTL. The top 5 and 10 percent of the highly connected genes were analyzed for over-representation of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (Kanehisa & Goto 2000) using the analysis tool, WebGestalt (Zhang et al. 2005), which performs a hyper-geometric test for the similarity of the list of highly connected genes and the members of curated pathways in the KEGG database. The resulting list of enriched categories of genes represents those processes, functions and molecular classes that are most involved in genetic variation in behavior.
