Two-stage genome-wide search for epistasis with implementation to Recombinant Inbred Lines (RIL) populations

Gene expression data (filter.genes.csv): Gene expression Arabidopsis thaliana data was downloaded from the TAIR database website (ftp://ftp.arabidopsis.org/home/tair/Microarrays/analyzed_data/affy_data_1436_10132005.zip). A sample of 211 RIL population individuals derived from a cross between two inbred accessions, Bayreuth-0 (Bay-0) and Shahdara (Sha), was used. Transcript (mRNA) levels were quantified using Affymetrix whole-genome microarrays with two replications (arrays) for each individual. The gene expression data was preprocessed by the Variance Stabilization Normalization method [31]. Traits with essentially no expression (15,566 traits) were filtered out by the EM algorithm for a mixture of univariate normal distributions (using the ‘mixtools’ R package), leaving 7244 traits for the analysis.