Issues with averaging for protein multimer

[ioanaapapa]

I've tried running the code for a streptavidin tetramer for 100 frames as a test (I was running the longer trajectories - 1000,2500,5000, and 10000 frames - on scarf but they seem to be still running after 5 days without any error messages) and I got this error at the "Calculating Entropy Values" stage when running with averaging for the tetramer. The code ran and I got results for just one streptavidin monomer and for the tetramer without averaging.

error.txt

[skfegan]

When the groups are larger than one molecule the averaging is not working for the conformational states. Ioana's strepativdin example was giving some negative values for the united atom level conformational entropy for a few residues. I think this is because the code is dividing by the number of frames to get from the count to the probability and not taking the size of the group into account.

[ioanaapapa]

For most terms, results with the averaging for N frames match results without averaging for 4N frames, which is expected. However, this isn't happening for the conformational entropy at the united atom level , which is always significantly higher with the averaging. At the moment, the comparison is for 1000 and 400 frames because I'm still waiting for the calculations with more than 1000 frames.

This seems consistent for all residues and all residue types.