Hello PyRosetta users,
Can you please advise me on how I should go about calculating atom-wise energy terms?
Should I simply sum pairwise energies for the nearest N-neighbours or is there a better way of doing this?
Thanks!
Category:
Post Situation:
You are talking about the `pyrosetta.toolbox.atom_pair_energy.etable_atom_pair_energies` function? I gave that a go and was happy with the results (used relative to other atoms) but there's a big caveat that a few weights are needed as the values are out of wack. I think that is why the functions docstring/doxygen annotation uses the word "lj_atr" as opposed to "fa_atr", but "fa_elec" is the same word, but the values are not only scaled, but shifted! Note differential treatment of intra and inter makes no difference.
I think the issue may lie with the fact that if you do
the values aren't zero, which is saying something. No idea what.
Probably worth mentioning that 'fa_intra_rep' has a different weight than 'fa_rep', while the rest of the intra values are zero for ref2015
Many thanks, this is anyway helpful!
I had a play with this again as I needed it.
To keep it simple I made a pose with a Ca++ and Cl- 3 Å apart (viz. below) and run the above function... And I am not sure why I did not spot that simply the e-table scores per atom are 2x and that in my codeblock I had weighted the atomic scores, but not the residual ones. So the `lj_rep` is totally `fa_rep` or `fa_intra_rep`.
The scores are the same for Ca and Cl residues, as there's nothing else and actually makes it obvious why I should have divided by 2 in the first place.
About the self with self score "mystery"... The wacky scores are just those one would get if there was an atom occupying the same space.