I want to do chip-seq for some histone e chromatin remodeling enzymes. I would like to known how many reads I need to have for a reliable result.
What matters is unique reads rather than total reads. ChIP-Seq runs that give 20-30 million unique reads will work quite well for most factors.
Agreed, it is more accurate to think in terms of uniquely mapped reads, and 20-30 millions should be good. However, it also depends on your system, size of the genome, type of signal expected (broad or sharp) and so on. You can have a look at the different metrics and guidelines published by the ENCODE and modENCODE projects to give you an idea (PMID: 22955991)
I have done some analysis in that respect. We had samples sequenced at a depth of 15, 30, 50 Mil and I tried to calculate the increase in unique reads over duplicates. The saturation is observed with about 30-35 Mil unique reads which capture about 90-90% of unique binding but still in not 100%. So, genome specifically mouse=2.8 GB, human=3.3 GB, this number is fine, you can scale up or down this depending upon the genome size to go further on. With the histone marks, you should definitely aim for min 60 mil.
The following paper in NAR discusses in depth about the requirements of sequencing depth for each of the Histone PTM:
http://nar.oxfordjournals.org/content/early/2014/03/05/nar.gku178.full
- Badges
- Science method