Hi,

DiffBind is great and a heat map of counting reads is very helpful (e.g. the attached figure). However, I would like to run a hierarchical clustering using other graphical tools that I can swap different groups. Therefore, I need each sample's affinity scores for each consensus peak. The best way is that (1) every row is a sample; (2) every column is a consensus peak with affinity scores. Could you tell me how to get this kind of data? Many thanks.

Best,

Gary    

Hi Gary-

Use dba.peakset() with bRetrieve=TRUE:

> scores <- dba.peakset(myDBA, bRetrieve=TRUE, DataType=DBA_DATA_FRAME)

This will return a data frame with a row for each consensus peak, the first three columns being chromosome, start, and end; and columns for each sample. (If you don't set DataType=DBA_DATA_FRAME, the data will be returned as a GRanges object). So to get what you're after you would want to strip the first three columns and transpose:

> scores <- t(scores[,-(1:3)])

Remember that you can easily change the scoring method without re-counting by setting peaks=NULL in dba.count():

> myDBA <- dba.count(myDBA, peaks=NULL, score=DBA_SCORE_READS)
> scores <- dba.peakset(myDBA, bRetrieve=TRUE, DataType=DBA_DATA_FRAME)
> scores <- t(scores[,-(1:3)])

Cheers-

Rory