search

dozmorovlab / HiCcompare

Joint normalization of two Hi-C matrices, visualization and detection of differential chromatin interactions. See multiHiCcompare for the analysis of multiple Hi-C matrices
https://dozmorovlab.github.io/HiCcompare/
Other
18 stars 3 forks source link

Advice on value for A #31

Closed TeiturAK closed 11 months ago

TeiturAK commented 11 months ago

Hi,

could you please advice on the selection of A from the filter_params plots? It looks quite different fromt the examples I've seen. I also attached the MD (all chromosomes) after selecting A = NA which was then automatically set to 7. Is the assymmetry from low to high abundance regions something can be expected between samples coming from two completely different tissue types?

image

image

TeiturAK commented 11 months ago

Updated question with more information

mdozmorov commented 11 months ago

Hi @TeiturAK, you may set A=10 if you want to minimize false positives. That's the minimum average expression (A) where False positives rate drops and True positives rate increases. The automatic selection of A=7 is also appropriate.

Your MD plot also looks good. The data appears to be deeply sequenced, balanced. The disbalance of up- and downregulated interactions at shorter distances is expected for different tissue types. I suggest focusing on the distance range up to ~100, interactions gained in one cell type appear the main driver of differences.

TeiturAK commented 11 months ago

Thank you very much for the help and feedback. Closing this.