Question about .cool file bias vector values in regions with 0 contacts

[kalavattam]

I have received some .cool files for my work and I want to ensure they are formatted correctly. How are regions with 0 contacts represented in the bias vector? Should they be NaN, 0, or some positive finite value? Thank you.

[nvictus]

Hi Kris! Regions with 0 contacts are normally masked out for matrix balancing (the algorithm would never converge if they were kept). Masked/filtered bins are normally encoded as NaN in the weight vector, which will "NaN-out" the corresponding row/column of the matrix when the weights are applied (B_ij = w_i * w_j * A_ij).