Hi, when use diffbind to check ChIP-seq data we need bam file. Do I need to use the bam file which already removed unmapped reads /duplicates/blacklist? And what will be the right range of FRiP for a successful experiment? Thanks.
Yes, it is a good idea to remove blacklisted reads before using DiffBind. I usually don't bother removing duplicates unless I have an unusually high percentage of them - particularly for factors with sharp peaks where some duplication is expected.
The FRiP really depends on your sample and how heavily your factor binds. Histone marks get up into the 20-30+% range sometimes if you have a really good enrichment - TFs I usually expect more in the ~15% range, but again, it really depends on your factor and how well your ChIP worked. A general rule I like myself is that I want at least 2x enrichment over my inputs. So if my input has 6% of reads in peaks, I'd want to see at least 12% of reads in peaks in my actual ChIP. But ideally it should be higher, it just depends on how lenient you can afford to be with your samples.
The DiffBind authors have another R package called ChIPQC that is quite good and can tell you the percentage of reads in blacklisted regions, duplication rate, in peaks, general enrichment rate, range of signal, etc. It also makes some nice figures and is easy to use.