search

ay-lab / fithic

Fit-Hi-C is a tool for assigning statistical confidence estimates to chromosomal contact maps produced by genome-wide genome architecture assays such as Hi-C.
MIT License
77 stars 16 forks source link

[Question] Does the read depth of the raw/iced matrix generated by HiCPro influence FitHiC2 loop calling? #72

Open Miracle-Yao opened 3 months ago

Miracle-Yao commented 3 months ago

Hi, @ay-lab

Here's a question I'd like to get your advice on.

FitHiC is able to use the raw/iced matrix file information generated by HiC-Pro as input to look for biologically significant chromatin interactions. The hicpro2fithic.py input file includes the raw matrix, bin index, as well as iced.matrix.bias. I would like to know what can be done at this time if I want to compare the LOOP between different samples, but the depth of the raw/iced matrix between the samples has a big difference. How is depth normalisation performed in this case, and if depth normalisation between samples is not performed, does this affect the truthfulness of the loop?

Thank you and I look foward to your reply. Yao Hua

ay-lab commented 3 months ago

hi Yao. Good question. FitHiC does not perform normalization for depth and its ability to discover loops depends a lot on the sequencing depth. One trivial but less than ideal solution is to downsample one of the samples to the depth of the other. For more proper ways of comparing, you should look into tools that do differential contact/loop analysis. There are many - none of them perfect.