Understanding enrichment signal from TF ChIP-seq using phantompeakqualtools and plotFingerprint(deeptools)

Question

[ahnje770]

Hello,

I have a question regarding the TF ChIPseq enrichment signal from this ChIPseq data.

Samples include:

1. Input Control ("INPUT_S17")
2. Untreated sample ("UNT_S18")
3. Treated sample ("TNF_S19")

Other information:

• The fastqc results show great quality reads
• mappability is good (over 90%)
• the number of read fragments after filtering for blacklisted regions, duplicates, and multimapping reads was 11 million, 37.2 mil and 26.7 mil for Input control, Untreated sample, Treated sample, respectively)

However, I'm a bit confused regarding ChIP enrichment signal as evaluated with the strand cross-correlation figure (from phantompeakqualtools) and plotFingerprint figure (from deeptools).

The strand cross correlation shows a strong signal in both untreated and treated samples. There also seems to be a moderate signal coming from the input control (shown below).

However, with the plotFingerprint figure (from deeptools), there doesn't seem to be much ChIP enrichment in both untreated and treated samples in comparison to the input sample. Would this indicate that the ChIP experiment is unsuccessful?

Running macs2 callpeak with the input sample as the control (-c parameter) gives: - 139 peaks in the treated sample (TNF_S19) - 6875 peaks in the untreated sample (UNT_S18)

However, when running macs2 without the input sample as the control I get: - 24 peaks in the input (INPUT_S17) - 16429 peaks in the treated sample (TNF_S19) - 28650 peaks in the untreated sample (UNT_S18)

[varamhcu]

Hello, I have a query similar to this post regarding the SPP peak calling result. Can you help me understand the result ?. Here I am attaching the link for the post (Negative strand shift value in cross-correlation plot (spp peak calling))

Thank you