Diffbind-weird Fold value

Question

[wei.zhang]

Hi all, I am new to Diffbind, when I used it to ran my ChIP-seq data, I got below result. From the dba.report if my understanding is correct Fold=Conc_GNP1-Conc_GNP2, but for my data is obviously not, anyone can help me with this? Thanks a lot!

result<-dba.report(Epigenome2,th=1)
result
GRanges object with 59140 ranges and 6 metadata columns:

seqnames            ranges strand |      Conc Conc_GNP1 Conc_GNP2         Fold     p-value       FDR
           <Rle>         <IRanges>  <Rle> | <numeric> <numeric> <numeric>    <numeric>   <numeric> <numeric>

chr20 16984425-16985425      * |   5.53053   5.85727   5.10743   0.00212363 0.000522680  0.999946

Answer 1

[Rory Stark]

In the latest version of DiffBind, the default is to use fold changes calculated by the underlying analysis package. The default analysis package DESeq2 does a shrinkage estimation of fold change; the value in the report reflects this value, and not a simple calculation based on the sample group concentrations.

[wei.zhang]

Thanks a lot Rory!!!