ENCODE ChIP-seq data

[hwu]

Hello,

I'm confused with some notations for ENCOE ChIP-seq data generated by Yale. If you go to http://genome.ucsc.edu/cgi-bin/hgTra...odeYaleChIPseq, what exactly is, for example, Pol2 (IgG-mus) factor? I understand IgG-mus is kind of non-specific antibody. But is Pol2 (IgG-mus) the data from Pol2 subtract Input (IgG-mus), or something else?

Thank you very much.

[sglandt]

The (IgG-mus) designation means that peaks for this dataset were called by the PeakSeq algorithm (see Rozowsky et.al., Nature Biotechnology 27(1) 66-74 for details) using an IP with mouse IgG as the control. For alot of the earlier data and all the PolII data not labeled with IgG-mus, input DNA was used as the scoring control

[hwu]

Thanks a lot for replying. So IgG-rab means using the IgG-rab as control, right? Which peaks are better, the one using input DNA or mouse igg?

[sglandt]

Its kind of a matter of preference and for really strong peaks it probably doesn't matter a great deal. Our group has switched to IgG as a control because it removes some very strong non-specific peaks. Input is used by many because it helps correct for sonication bias which does affect peak calling. However, sonication efficiency correlates well with open chromatin and regions of open chromatin are, not surprisingly, the places where transcription factors tend to bind. Therefore, by using input as a control, you are somewhat preferentially biasing your scoring against some real peaks.
This is discussed in Auerbach et.al. Proc Natl Acad Sci U S A. 2009 Sep 1;106(35):14926-31.

[hwu]

This is great. Thanks for the answers!

[Subhajyoti]

Overlap of Chip-seq peaks from multiple cell-lines

How can I identity the Chip-seq peaks (for a given TF) that are shared across most or all of the ENCODE cell-lines. One option could be to use ENCODE Chip-seq peak calls and apply intersectBed to find overlaps, but the resulting dataset might get smaller and highly fragmented with an increase in the number of cell-lines. Any other suggestions?