search

josura / c2c-sepia

Multi agent model for perturbations in networks of networks. Used for Cell-to-cell communication with enriched pathways, using scRNA-seq or spatial data
Creative Commons Attribution 4.0 International
2 stars 0 forks source link

Normalization in the W and W* matrix #14

Open josura opened 1 year ago

josura commented 1 year ago

I think that the normalization done with the whole summing up of absolute values is a little too much, maybe some shrinking to the normalization factor could be better probably, even though the overall results should be kind of the same.

The shrinking method could be to count the backward edges from the node itself and use this number to shrink the sum of absolute values

josura commented 1 year ago

Also, I think that the whole normalization should only take into account the number of edges, not their values, since some of them could be negative (and right now it is ambiguous since the values of the edges are all 0, -1, or 1)