HOMER tool's HiC interactions

Question

[hiren.knowledge]

Hi,

I do have a question on finding significant HiC interactions using HOMER tools.

When I use HOMER tool's "analyzeHiC" program to capture significant HiC interactions at P <= 0.001 cutoff, in the output file at the column # 19 it provide FDR (calculated by Benjamini method with no. of analysed tests i.e., 400 million tests) values for all the significant interactions.

The FDR value is always between 0 and 1 and maximum (~90%) of the HiC interactions are with the FDR value "1". Does it mean that all these interactions are significant but with false discovery rate 100%?

Should I consider HiC interactions, found with P <= 0.001, significant ?? OR Should I also consider these strange value of FDR?

Thanks in advance for helping me.

Hiren

Answer 1

[Sukhi Singh]

In a different scenario (e.g., significant binding sites from ChIP-Seq), I would not use the peaks above a certain FDR (<10-20%). I am not sure how the background model is calculated, but if you go in depth with that you would could speculate why the FDRs are so high. Best would to contact Chris Benner itself, on how the FDR are computed if the following page doesn't answers.

http://homer.salk.edu/homer/interactions/HiCBackground.html

[hiren.knowledge]

Thank you Sukhdeep Singh for your valuable response. I have sorted out the problem by considering 10% FDR cutoff and selected the HiC interactions.

Thanks,
Hiren