Minimum Coverage for ChIP-Seq & RNA-Seq

[canis]

Hi all,

Can anyone please suggest what should be the minimum Coverage and Read Depth required to get meaningful results for ChIP-Seq and RNA-Seq analyses.

Thanks,
Veer

[SGPSling]

I don't think this question has been examined carefully to date. You may want to read the recent paper of Rozowsky et al. Nat. Biotech 27 (1) 66-75. The bottom line is different transcription factor has different saturation depth. We simply need ChIP seq data of more transcription factor to see if there is a generalized pattern for different class of TFs.

Anyone can add more reference discussed the same matter will be really appreciated.

[canis]

Thanks SGPSling I will go through the paper Rozowsky et al. Nat. Biotech 27 (1) 66-75. My understanding of what you have posted is that the saturation depth can not be predicted in advance , it will depend on the sample used. Am I correct in my understanding?

[james hadfield]

SGPSling is correct but I would say 1 lane for ChIP-Seq and 2-3 for mRNA-Seq are good ballparks.

[SGPSling]

Hi, Canis: I would say the saturation depth depends on the TF you are studying. In the paper I quoted, the identified peaks saturated arount 3-4 M reads for PolII so that one lane would be enough. But for STAT1, it didn't reach plateau even when they sequenced 24 M reads.

For most of the ChIP seq papers published so far, they usually got 1-2 M reads. So James is right that people usually do one lane for each sample in a ChIP seq experiment at this moment.

As for now, we cannot predict the saturation depth for ChIP. But I believe with more results published in the future, one migth be able to do so. At this stage, we simply need more data to improve our knowledge.

[SGPSling]

Sorry, don't know where the thumbs down icon come from in my last post. It is not what I want.

[ECO]

Sorry, don't know where the thumbs down icon come from in my last post. It is not what I want.

Fixed!

[james hadfield]

I said one lanes should do for ChIP-Seq and that comment is based on us geting up to 20M reads from a single lane. We routinely get over 10M reads and (I am a little gobsmacked by this) saw a sinlge control lane generate 0.97Gbp from a 42 cycle run!
If you only get 1-2M reads then you need more lanes!
James.