I would like to use information stored in the DNase bam files to support my predictions but looking at the data, it appears that called peaks and reads don't always match up. For example, there is only one bam file for H1-hESC and some of the DNase peaks called for H1-hESC have read distributions with <5 reads (e.g. chr4:139,247,866-139,249,698). This doesn't seem like enough support to call a peak. I've also observed similar occurrences in GM12878, which again only has a single bam file. Are there any bam files missing? I understand the other cell types with multiple technical replicates were merged before peak calling but in the case of H1-hESC and GM12878, are these the bam files that were used to call peaks for DNase? Is it possible that so few reads can provide enough evidence to confidently call a peak?
Created by Alex Essebier aesseb Alex,
Please see https://www.synapse.org/#!Synapse:syn6131484/discussion/threadId=627 with the corrections.
Thanks again for catching this and apologies for the errors.
-Anshul. Alex,
We haven't forgotten about this. We'll be sending out a message about the BAM file additions later today.
Thanks,
Anshul. No inconvenience here, just glad there's an explanation!
Thanks for fixing it up.
Alex Alex,
Thanks for catching this. We did a check on all the BAM files and indeed, there were some replicate mismatches between the BAMs provided and the ones we used for the peaks/signals. To clarify, the cell types all match but there were some replicate mismatches between what was used for the peaks/signals and the BAM alignment files we provided. We are just replacing these right now and will post a detailed update on files that were added, replaced and removed to the message board and the wiki itself. Note this only affects the BAMs. The peaks and signal tracks are fine. Apologies for this error. Hope it didn't cause you too much inconvenience.
Thanks,
Anshul. We are looking into it. We intended to provide all BAM files that were used to generate signal tracks and call peaks. Will check and get back to you tomorrow.