ChIP-Seq / Peaks Caller

[nguyendofx]

Hello All,

There are many ChIP-Seq Peak detection applications available (open source and commercial). I don't know which one better than other AND when to select a right one?

The short list: MACS, QuEST, SISSRS, HOMER, ERANGE.

Which one is good for peaks finding (E.g Transcription Factor binding, histone, polymerases ..etc)

Thanks,
Ng

[mudshark]

As far as I know there has not been ANY objective comparison of Peak callers on a sufficient number of non-virtual reference data sets (having different read depths, different binding modes, different genome sizes, different mappability fractions..) to estimate their performance. There are some papers out, however..

furthermore, it might be of interest if the peak caller is actually able to make a two-sample analysis (ip versus control) and if it can operate on replicate samples.

if you are interested in my experience - for sure not universally valid - but tested on quite a lot of complementary data with different sequencing depth:
QuEST > SISSSR > Cisgenome2 > MACS in terms of specificity
SISSR > MACS > Cisgenome2 > QuEST in terms of sensitivity

unfortunately Quest and SISSR cannot deal with replicates, MACS?..