search

deeptools / HiCExplorer

HiCExplorer is a powerful and easy to use set of tools to process, normalize and visualize Hi-C data.
https://hicexplorer.readthedocs.org
GNU General Public License v3.0
231 stars 70 forks source link

how to choose correct PC #656

Closed BenxiaHu closed 3 years ago

BenxiaHu commented 3 years ago

Hi, I am runnung HicPCA to analyze compartment analysis. Would you have any good suggestions to choose the coorect PC for each HiC? Thanks in advance. Best,

joachimwolff commented 3 years ago

The correct PC is usually PC1 or 2, in rare cases it is PC3. A simple way is to see if it matches the checkerboard pattern of the Pearson matrix.

gtrichard commented 3 years ago

To complement Joachim's answer, be sure to check that the PCs are correctly oriented, i.e. that positive values are corresponding to active regions and negative values to inactive regions. You can determine that by calculating genes density, GC content, average binned genes transcription, active histone marks or open chromatin.

"Usually" the correct PC is the one showing ups and downs in short succession. Different PC can be the "good" one for each chr (for instance PC1 for Chr1 and PC2 for Chr2). You can although try to get the correct PC values as PC1 for each chromosome by for example masking certain regions such as large inactive regions at the beginning or end of chromosomes (at least in Drosophila).

If your samples have low compartmentalization, it is recommended to call A/B compartments in sub-matrices of your chromosomes (for instance call A/B comps in chr1:0-10,000,000, chr1:10,000,000-20,000,000, etc) and stitch the results and compare them with the checkerboard pattern of the matrix.