I have ChIP-seq data for samples at three different time points, and the last time point has almost twice the number of reads as the other two. A repeat of the ChIP-seq does not show this pattern. Why would one sample have so many reads? Does this make my first data set unusable?
It depends- 1) is the coverage/depth of sequencing is different in 3rd case? 2) There could be PCR duplicates more so better filter the PCR duplicates ( use samtools or picard and many such tools/utilities available) then see the reads. 3) or may be something is not sequenced in first two samples ( all Chroms etc). Only after seeing the fastq files someone can tell the exactly why its so.
If the RAW number of reads generated in the third experiment is higher, it probably simply means that more sample has been loaded in the NGS instrument -> more clusters (assuming Illumina) have been generated and more sequences produced. Sensu stricto, simply counting raw (absolute) read number has no biological meaning. Just to compare it with something more 'classical': trying to get a biological meaning from raw read counts is like trying to analyze the expression of a gene with Q-PCR in absence of an housekeeping.
first, did you calibrate the library size? By comparing difference samples you need to make two libaries the same size. Second, did you remove the duplicated reads? For ChIP seq we typcially collapse all identical reads to one. If you use some tools like MACS to call peaks, it will calculate the actual RPM for each peak, I remember.
Although the technicians will calibrate each libraries into the same amount during the sequencing process, it always has variations. We saw the many times when received the seq data. One sample has about a lot more reads than others.
Many thanks to everyone for your replies. I am currently investigating all the possibilities and trying to find the best method of analysis. Best.
- Badges
- Science topic
- Similar topics
- Gels