Hello everyone,
I want to ask a general question and your valuable comments about my approach.
I am analyzing different ChIP-seq samples from different datasets under same conditions (i.e. : mcf7 er alpha control chip-seq under the same conditions). Some of them have really high peak numbers or contrarily very low compared to the numbers they should normally have.
I thought that I can compare my files with the same samples in ChIP-atlas. I see that also in ChIP-atlas, peak numbers could be really high. But I am not sure if it is making sense or not. If you could comment I would appreciate it. Thank you.
It is generally a good idea to obtain some basic QC metrics about the individual ChIP-seq samples at hand. One of the most informative metrics IMO is the "fingerprint", which tells you about (a) the coverage of your sample, (b) the type of enrichment, (c) the strength of enrichment and (d) how similar to an input sample the individual samples are. You can find more information in the docs of deepTools.
The ENCODE consortium also recommends to look at the fraction of reads in peaks (FRiP) and other scores. The CISTROME DB paper has come up with some thresholds that they employ to determine whether a TF ChIP-seq meets their QC criteria, I'll paste the table here:
Thank you for the information and recommendations. It is very helpful for me. I actually have chosen all of my samples from Cistrome - and the ones passing all of the quality control.
Log in to answer this question.
You should include your code for peak calling and the results you are getting, otherwise we can't really give any advice.