Hi Rosetta community,
I'm running the prepack protocol using as input a PDB file containing a dimer (chain A and B). The protocol seemed to work ("... reported success in 3 seconds") but I get the following warning :
[ WARNING ] When switching to a fa_standard ResidueTypeSet: Pose already contains fa_standard ResidueTypes.
and my score_sc file contains 0.000 for the total score for every generated structure.
Here the command I used to run prepack:
$Programs/rosetta_src_2017.13.59376_bundle/main/source/bin/docking_prepack_protocol.mpi.macosclangrelease -s dimer_BW.pdb -docking:partners A_B -docking::sc_min -nstruct 4
My pdb file contains no header and all heteroatoms have been removed.
Am I doing something wrong ? I can't really see the problem.
That's nothing to worry about.
As the final step, the Docking Prepack protocol transforms the structure into fullatom mode. Since in your case the structure is already in full atom mode, the type switching code pops up a warning. -- In your case you probably *do* want to have the structure in fullatom throughout, so that's nothing you should be concerned about.
The missing total score is also a consequence of that residue type set mode switch. As part of the switching code, all the energies in the pose are dumped, as generally speaking the energies of one mode aren't comparable with another. As there isn't any additional scoring step after this mode switch, the energies stay empty, and you get a zero total score reported. Don't worry, the structure should be prepacked appropriately anyway.
Thank you for the answer.
Indeed, the input structure is in full atom.
I previously ran a different build of Rosetta3.4 on the same dimer complex (in full atom as well) and the output was completely different. I had the dimer_0001.pdb, away_dimer_0001.pdb and away_packed_dimer_0001.pdb files as output structures, also the score_sc file was filled with the different energy terms:
SCORE: total_score Eaway Eaway_packed Einitial Eprepack dslf_ca_dih dslf_cs_ang dslf_ss_dih dslf_ss_dst fa_atr fa_dun fa_pair fa_rep fa_sol hack_elec hbond_bb_sc hbond_lr_bb hbond_sc hbond_sr_bb description
SCORE: 3258.624 -401.257 -496.735 -117.559 3258.624 0.202 0.201 0.066 -3.656 -940.754 24.538 -8.117 3917.073 398.977 -3.550 -1.962 -2.762 -2.695 -118.936 dimer_0001
In my current run as I said, the score file reported a total score of zero and I obtained only 2 output structures, dimer_0001.pdb and dimer.prepack.pdb (same files). As you said, the prepack protocol might have worked, since the output of this run is identical to the one I've obtained with my previous run. So in the following case, am I expecting the away_... and away_packed_... structures?
"As part of the switching code, all the energies in the pose are dumped, as generally speaking the energies of one mode aren't comparable with another. As there isn't any additional scoring step after this mode switch, the energies stay empty, and you get a zero total score reported."
By this, you mean that I wasn't supposed to get scores in my previous run as well.
The forced switch to full atom is a relatively recent addtion, so would not have happened in Rosetta 3.4, leaving you with a scored output. -- In your case the switch is from full atom to full atom, so the energies are comparable. The type set switching code doesn't really catch onto that, though, so it removes the energies regardless. (I just got off an email from the developers, and they agree that there's a better way of handling this, and this behavior will be going away in the future.)
The removal of the "away" structures is another recent change. In practice there's not much use of the away structures, and they can really clutter up the directory, especially in some of the ensemble docking protocol. For that reason, the code in main/source/src/protocols/docking/DockingPrepackProtocol.cc which outputs the "away" and "away_packed" structures has been commented out. If you really, really want them, you can certainly uncomment those lines and recompile, but I doubt that they would actually be all that much use for you.
True, I don't need the "away" and "away_packed" structures. But, I was wondering why using the same protocol gave differen outputs.
Thanks a lot.