How reproducible is TF ChIP-Seq?

[Livi81]

Has anyone looked at the reproducibilty of ChIP-Seq for transcription factors?
I realise that in the past expense has meant that many papers have been published without duplicates. However as costs have begun to fall has anyone seen duplicates being done and if so to what extent do peaks overlap between samples?

Any input welcome.

[Chipper]

It is discussed in the PeakSeq paper (Park et al) and the standard for Encode datasets are now to include two replicates.

[ETHANol]

There are quite a few papers out there with no replicates and/or without input controls. Even with the high cost of sequencing, I don't think it was an excuse. I'd bet pressure to publish is more of an issue then cost most of the time.

I'd hope that anyone reviewing papers from here out requires a minimum of two biological replicates of both ChIP'ed DNA and input DNA.

I think another issue is how are reviewers to assess data when they are given perhaps a couple cherry picked peaks in a figure and not browsable tracks?

The bottom line is reproducibility varies. Variation will go down as techniques, expertise and technology improves.

[Chipper]

If you need replicates or not depends on the type of questions you are asking and in what other ways you follow up the results. Also, in most cases the "single replicates" were actually several pooled ChIPs.

[Livi81]

Ok thanks for the replies, much appreciated.
I'll check out that reference.