You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I did some looking and folks don't seem to be happy with how the force-directed layouts in igraph handle negative weights. Often the examples i saw only had a few nodes, so i think we should see how it goes with our networks and modify if needed.
If we do need to modify, one fun idea would be to have options for force-directed: positive only, negative only, both, or absolute value. Basically ways to keep all the edge weights on one side of 0 (except the "both" option). Kathryn was talking about how she's actually only interested in negative edges for the wgcna stuff, which led me to think about this plan.
The text was updated successfully, but these errors were encountered:
I don't think any of the layout methods in igraph support negative weights. If they don't show an error when you pass a negative weights, that is likely a bug that we should fix.
If you want to discuss why a specific layout does not support negative weights, and whether support would make sense, you can post on https://igraph.discourse.group/ or ping me here.
If you find that igraph accepts negative weights, but shouldn't (e.g. produces bad output), please open an issue in the igraph repo.
I did some looking and folks don't seem to be happy with how the force-directed layouts in igraph handle negative weights. Often the examples i saw only had a few nodes, so i think we should see how it goes with our networks and modify if needed.
If we do need to modify, one fun idea would be to have options for force-directed: positive only, negative only, both, or absolute value. Basically ways to keep all the edge weights on one side of 0 (except the "both" option). Kathryn was talking about how she's actually only interested in negative edges for the wgcna stuff, which led me to think about this plan.
The text was updated successfully, but these errors were encountered: