CouplingsScoresCompared_longrange.csv misses couplings

[ddingding]

Hi,

Just to mark that for the complex pipeline, the CouplingsScoresCompared_longrange.csv is missing some inter couplings in the complex when the couplings are close (for example a coupling A_i,i=60 with A_j,j=58 is missing in my protein complex but this coupling is indeed displayed on the contactmap).

Presumably the filter for longrange is not taking into account that the 2 positions i and j are coming from different proteins.

Cheers

[kpgbrock]

Hi, which version of evcouplings did you use (or which environment on o2?) I saw this bug in a previous version, but I think it's been corrected in the latest - let me know and I'll follow up on if we need to update code/envs!

[ddingding]

Hi,
Yeah I'm using the evcouplings_dev_test on o2.

These wrong short-range inters are also missing from the CouplingScoresCompared_all.csv file, but they are in the couplingsScoresCompared_inter file.

Cheers

[thomashopf]

Looping in @aggreen

[aggreen]

Strange yeah, I believe this is fixed in my new version and will check on it when I update that pull request