I have a problem with motifs. It is supposed that average binding is correlated with motif conservation but my ChiP-seq experement doesn't confirm it.
I called peaks (MACS 2), discovered motifs for them using MEME-ChiP and got average coverage of peaks from bedgraph files.
Then compared the best motif to all peaks' sequences using matchPWM from Bioconductor package and got scores. The peaks were divided into 4 groups according to their similarity to this motif from weakest similarity to strongest similarity (ranks from 1 to 4). And then I built a boxplot with average coverage vs. rank. The mean average coverage of peaks was almost the same in all groups! And no upward trend (I supposed that the more is similarity of group to a motif, the more should be average coverage). I repeated this for all motifs and got very similar results - mean of average coverages was almost the same. Moreover, I repeated it for another ChiP-seq experiment.. All the same - it is very strange.
Can please anyone tell me where I possibly made a mistake.