What do you do when a peak overlaps a gene? Do you choose these gen as a target?
The question is unclear. CHiP-Seq in principle identifies binding sites across the entire genome for whatever protein was immunoprecipitated. If you are asking which of the various identified binding sites should be followed up with further work, that depends completely on the specifics of the experiment and can't really be answered in general (IMHO). If you are asking what are the criteria for calling a "hit", my advice would be to read papers in the literature for the technical details and see what cutoff parameters are generally accepted for your particular sequencing platform.
A peak is also only relevant if you have an input control, that is, you sequence fragmented genomic DNA in parallell with your ChIP sample(s). There is a lot of bias in what portions of a genome that is sequenced due to GC content etc, so even sequencing genomic DNA you will get peaks. Your ChiP-seq must be normalized to the relevant input DNA in order for any peaks to be considered to be true (due to specific interactions between your DNA-binding protein and the antibody).
For "old-fasion" ChIP experiments (assessed by qPCR not seq) you would also do a negative control with only beads or beads coated with bulk ("nonspecific") antibodies to control for proteins/DNA adhering to the beads in an antibody-nonspecific manner. In ChIP-seq experiments this is apparently not generally not deemed necessary. But You definitely need to sequence the input DNA! Only then can you start pondering how to define your thresholds.
Best of luck
It is also wise to combine ChIP-seq with expression profiling/microarray data. If the gene in question is up or down-regulated in the mutant of your transcription factor, I would be especially encouraged that it is indeed a genuine target. Verifying binding sites by DNase-1 footprinting and the establishment of a "consensus" may also help to further verify targets. Ultimately all in vivo-determined targets will range from high-probability to low-probability (based on the ratio of enrichment from IP/non-IP samples).
It is also worth emphasising the need for controls to remove false positives.
1. Total DNA (non-IP). This very useful additional control enables non-uniform shearing of the chromosome to be taken into account. More relaxed regions of DNA shear more readily thancompact regions, and so the former can give rise to higher background signals in ChIP-seq (Reference attached Teytelman L, Ozaydin B, Zill O, Lefrançois P, Snyder M, Rine J, Eisen MB. 2009. Impact of chromatin structures on DNA processing for genomic analyses. PLoS One 4:e6700. doi:10.1371/journal.pone.0006700.)
2. There has evidence that highly expressed genes (hyper-ChIPable )are more subject to false positives and so some people recommend including a control of a FLAG-tagged protein that is not a transcription factor (although this has not yet been widely adopted)
- Badges
- Science topic
- Similar topics
- Biological Science
- Genetics
- Genetic Analysis