residue_energy_breakdown of a protein-ligand complex.

2 posts / 0 new
residue_energy_breakdown of a protein-ligand complex.
#1

Dear Rosetta Users,

I'm using the "residue_energy_breakdown" utility in order to obtain a per-residue energy breakdown of the residues that interact with the ligand in my protein-ligand complex generated by the RosettaLigand application.

In the POSE_ENERGIES_TABLE at the end of the docked model I read quite reasonable energy terms :

fa_atr -2382.64
if_X_fa_atr -24.0332
fa_intra_rep 4.83652
fa_rep 235.344
if_X_fa_intra_rep 0
if_X_fa_rep 3.43482
hbond_bb_sc -123.179
hbond_lr_bb -300.454
hbond_sc -70.8115
hbond_sr_bb -444.803
if_X_hbond_bb_sc -2.05341
if_X_hbond_lr_bb 0
if_X_hbond_sc -5.84876
if_X_hbond_sr_bb 0

but when I try to breakdown the energy on a per-residue basis, I get these values for the residue-ligand interaction :

fa_atr -8.196
fa_rep 809.474

fa_atr -3.675
fa_rep 14.476

fa_atr -8.580
fa_rep 26.700

I can't figure out why the Lennard-Jones repulsive force is so high. There are no evident clashes in the .pdb input.

I'm trying to relax the protein-ligand complex by using the default relax utility or the all-atom harmonic-restrained relax protocol (even though I think that it should be well suited for the Rosetta "residue_energy_breakdown" utility, since it has been generated in the RosettaLigand scorefunction), but I'm not sure how to handle the ligand, as the ligand bonds become totally distorted.

Samuele

Category:
Post Situation:
Tue, 2021-05-18 04:45
sam_dc

You don't necessarily need a large clash to get a high fa_rep score. The score increases rapidly once you get the atoms in too close of proximity. That's consistent with the distortion on relax, as well. Since the fa_rep score is very high, relax is willing to do awful things to the ligand to lower it.

The residue_energy_breakdown application should output interactions in a pairwise fashion, so you should be able to tell if the high fa_rep score for the ligand is coming from interactions with particular residues. I'm guessing that it's coming from just a few residues, and if you do measurements, you'll find that the atomistic approach is closer than you normally see with ligands. (Don't forget to include hydrogens - they contribute to fa_rep as well.) Those residues will also likely be where relax is "crushing" the ligand, distorting it to get away from those overlaps.

You're completely right that the all-atom harmonic-restrained relax protocol is normally the way to correct this. If you are sure to include the -ramp_contraints false flag, it should keep the atoms with less than an Angstrom of their starting position, while removing the bulk of the repulsion.

One caveat, though, is that you probably want to make sure you're using the same energy function flags for your manipulations. That is, if you're interested in the ligand scorefunction, run both residue_energy_breakdown and relax with that scorefunction, and don't forget the -restore_pre_talaris_2013_behavior flag that goes along with that. When pre-relaxing, it helps to have the same scorefunction for prerelax as analaysis, as different scorefunctions have slightly different behaviors with respect to where their atom overlaps start to take off.

Tue, 2021-05-18 07:21
rmoretti