16 months ago by
CRUK, Cambridge, UK
I'm not sure what you mean by "standard scores." Do you mean a score generated by the original peak caller [eg -10 log10(pvalue)]? Or a statistic relating to the differential binding itself (ie FDR, fold change, mean concentration, etc)?
If you want to retrieve scores generated by the peak caller, there are a couple of complications. One issue is that the consensus peaks do not map directly to called peaks. The peak intervals may be expanded during the peak merging process, and not all peaks are called for all samples (so there is no score from the peak caller). Another issue is that
DiffBind normalizes the original peak scores to a 0..1 range, so the actual original scores are no longer available.
Let me know if you are trying to retrieve the peak caller scores, as I could give you some sample code. The basic approach is to read the original peak calls with their scores into a
GRanges object, and then overlap each one with the
GRanges object returned by
dba.report(). You can then use the original scores from peaks that overlap the significantly differential bound regions.