Entering edit mode
23 months ago
a_bis
▴
40
Hi all, I'm trying to generate a heatmap of my POI-bound genomic regions, which I've identified by CUT&RUN, and I'm finding that the output I'm getting is mostly 'empty'. Looking at my bigwig file, every peak that has been called seems to show a strong signal relative to the IgG control, and yet the heatmap is looking pretty empty (see images below). Any ideas on whether the files could be processed differently so I can get a better result? I'm fairly certain the heatmap output isn't reflective of the biology. As it stands, I have carried out the following:
computeMatrix reference-point -S <bigwig.bw> -R <peakfile.bed> -o sample-S1 --outFileNameMatrix sample-S1-matrix.tab --referencePoint center -a 500 -b 500 --sortRegions descend --sortUsing region_length -p 10
plotHeatmap -m sample-S1 -o sample-S1-heatmap.svg
Any suggestions will be very welcome, thanks!
To clarify, the track shows a typical called peak -- the three first tracks are my experimental replicates and the fourth one is an IgG control replicate.
does your peak file list a peak for all genes or just those called? You might have to filter your
peakfile.bedfile to only include genes that have an actual peak.Just those called! It's under 1000 peaks, I just checked the wc -l again to be sure.
What does the distribution of peak p-values look like? (or peak enrichment scores). Does the distribution (histogram or density plot) have a peak followed by a long tail? That would indicate some collection of strong peaks (easily heatmap viewable) and many smaller peaks that would be harder to see, or look flat in a heatmap.
These are the distributions for signalValue, pValue (-log10), and qValue (-log10) from my peak file.
Taking into account what you're saying, then, I'm now less concerned about the lots of 'blank space' and more concerned that my IgG control heatmap looks so similar to my sample's.
I just repeated the heatmap generation process with my IgG control and got an identical heatmap... I can't really explain it, I'm pretty sure my peak file looks fine and contains my regions of interest and only those.