plgem

The plgem software tool addresses the identification of differentially expressed genes (DEG) in high-density oligonucleotide microarray data, considering the intrinsic measurement variability. Leveraging an empirical power law global error model (PLGEM), the tool empirically models the dependence of variance on the mean expression level. This modeling is based on both in-house and publicly available datasets, demonstrating that the dispersion of repeated measures follows a power law dependent on the location of the measures. The proposed DEG identification method utilizes a statistic that explicitly incorporates model-derived estimates of measurement spread, coupled with a resampling-based hypothesis testing algorithm.