Statistical significance for discriminative power differences between the predictive methods.

The table contains the Benjamini-Hochberg corrected -values of the differences (using a Wilcoxon signed rank test) between each pair of methods. 1st-order and detailed TFFMs are likely to perform similarly and so between PWM and DWM. Results are obtained using data sets for which at least one method obtains an % when discriminating ChIP-seq data from genomic background sequences. Results obtained with HMM-generated background sequences are given in Table S2.