Hello,
I have ChIP-Seq data stemming from 3 samples, each with 3 replicates and 1 input sample. I ran the ENCODE ChIP-Seq pipeline. After annotation of the reproducible peaks that is generated by IDR, I am trying to do some gene set enrichment analysis on these genes. I have tried to use the following strategies for the same:
1) GSEA: According to the FAQ section of GSEA, ChIP-seq data can be analyzed using the preranked GSEA module however I do not have a huge number gene list that are ranked. I just have a bunch of genes that are ranked by the q value that IDR generates. So there was no particular result that I found here.
2) Chipenrich: I used the hybridenrich function of this software as I have few peaks but there were significant terms that are enriched.
3) clusterProfiler: neither enrichKEGG or enrichGO gave me anything significant either.
Is there a better approach to what I am trying to accomplish?
Thank you, Asma
Can you elaborate on what are you trying to accomplish? Do you have a list of peaks thresholded (peak / not a peak), or do you have a list of peaks ranked with p-values with no threshold? What kinds of pathways are you looking for: regulons, GO pathways, genome-wide correlations, etc?
I have 3 replicates per sample and 1 input sample each. I want to somehow combine these three replicates of each sample to get the FDR and P-values to do the ranking list. Hence I chose to run ENCODE ChIPSeq pipeline in "TF" mode as it runs IDR at the end.
This pipeline uses spp to call peaks per sample per replicate. After which IDR uses these ranked lists of peaks (not thresholded) to separate signal from noise based on confidence of rank consistency and reproducibility of identifications called IDR threshold. So ultimately I have these (IDR) thresholded peaks which I later annotate using ChIPSeeker. It is these annotated genes that I want to do some pathway analysis with. I am interested in GO, KEGG, genome-wide correlations.
I see; of the top of my head, I don't know any gene enrichment analysis tools that take replicates into account. Maybe you could use something like StereoGene (link) directly on your peaks? Now, if you want to do the search on gene level: in our lab, we use our own tool called iPAGE because it takes continuous p-values or other scores associated with the input genes, you might consider that if you want to use weak threshold (link). If you want to just go with the simplistic gene enrichment analysis, I find g:Profiler (link) very convenient