Hello,

I want to create a ChIP heatmap using profileplyr, but I want to merge some samples according to condition. This is implemented in DiffBind by the looks of things, but I want to customise the heatmap a bit to show the conserved peaks, gained peaks, and lost peaks in the same plot (from what I can tell, DiffBind only shows gains and losses). I also want to be able to combine all samples into a single heatmap.

Is there a way to do this in Diffbind, or directly in profileplyr?

Many thanks, Jess

You can definitely include three sets of peaks on a profile plot in DiffBind, using the samples parameter. There are a number of ways to do this, either by passing the indices in a report, or using a results-based report.

Note that peaks that are not identified as gained or lost are not necessarily "conserved"; higher FDR values mean only that there was insufficient evidence to identify them as changed, not that there was evidence that they are unchanged. Combining high FDRs with low absolute fold changes could give you a good candidate list of "conserved" sites, and indeed a profile plot would be a good way to visually verify that this is the case.