cmdoret/hicreppy

Do we need to drop first diagonals?

Closed this issue · 1 comments

Hi!
Could you clarify whether the method is sensitive to non-deleted zero and first diagonals of Hi-C map. I noted that they are taken into account by computing SCC. Will it give us the correct result?

Thank you in advance!

Hello @shappiron,

By non-deleted zeros, do you mean zeros that are explicitely stored in the sparse matrix inside the cool file ?
If so, it should not change anything as the matrix gets converted to COO format in both htrain and scc commands, which erases explicit zeros.

The first diagonals are taken into account, yes. Are they excluded in the original implementation ? If so, I missed that.