Significance analysis and improved discovery of disease-specific Differentially Co-expressed Gene Sets in microarray data

Kostka and Spang proposed a statistic (KS-statistic) and an algorithm (KS algorithm) to elicit Differentially Co-expressed Gene Sets (DCEGSs) by minimising KS-statistic. We prove that the statistical distributions of KS-statistic under null hypothesis in variance un-normalised and normalised data settings are central and doubly non-central F-distributions, respectively. Based on this analysis, we propose two alternative but equivalent statistics whose null distributions are easier to evaluate. Further, we propose to improve the algorithm by objectively setting the search parameters via maximising the statistical significance of the resultant gene set and pre-filtering the genes by Friendly Neighbours (FNs) algorithm.