I'm new in ChiP-seq analysis. I'm using SICER to analyze histone modification data with a control library. I have some doubts regarding the output file. In the output file (summary of filtered islands with an FDR < 0,01) there are some peaks which appear to be more enriched in the control library respect to the chip one. Is it normal? I thought I would have peaks only enriched in chip library. How is the fold change value calculated? Does it exist a best used threshold to filter this value?
General behaviour is we see control samples are not much enriched when compared to the ChiP peak regions. Ensure that you have equal number of replicates (biological replicate if you are merging them before peak detection). If you have unequal biological replicates:
For example: Control sample has two biological replicates and Test sample has single replicate then it may cause such differences. So, in this case try to run peak detection by considering individual biological replicate.
In such case, then check the FDR values for the enriched peaks in the control samples.
General way of Fold change calculation is by: Control Sample divided by ChIP Sample (or) vice versa
You might want to look into SICER's new incarnation in EPIC. You will find it much more accessible. And the creator is very responsive to issues on GitHub and answering questions if they are posted here or on Stackoverflow.
