search

tariks / peakachu

Genome-wide contact analysis using sklearn
MIT License
57 stars 9 forks source link

Question regarding fig. 2 in your paper #17

Open zhyanlin opened 2 years ago

zhyanlin commented 2 years ago

Hello,

I have a question regarding figure 2 in your paper. The figure said the recalls are 60%~80% when using ChIA-PET or HiChIP data as target. But peakachu's prediction is around 15K, and the number of peaks in CTCF ChIA-PET is almost 100K. Did you pool ChIA-PET as well?

tariks commented 2 years ago

Yes, we pooled ChIA-PET coordinates and allowed for a +/- 1 bin tolerance when determining matches.

zhyanlin commented 2 years ago

Thanks. It makes more sense!

How did you pool the ChIA-PET data? When I tried to use the peakachu's pooling tool to pool ChIP-PET, I got 0 loops.

I covert ChIA-PET data to the input of pool tool as follows: chr9 99230000 99235000 chr9 100010000 100015000 --> chr9 99230000 99235000 chr9 100010000 100015000 1 IF

Thanks!