- This topic has 3 replies, 2 voices, and was last updated 10 years, 6 months ago by Anonymous.
May 5, 2013 at 2:10 pm #1570Anonymous
When I were preparing some PDB files of crystal structures for Rosetta simulation, there are several ways to go after I checked Rosetta manuals and demos.
First method is following the doc ‘Documentation for relax pdb with all-atom constraints / prepare pdb for rosetta with relax’ in user guide and the codes in demos/prepare_pdb_for_rosetta_with_relax/. Breifly, it is using “relax” with some extra options e.g. ‘-relax:constrain_relax_to_start_coords’.
Another method is mentioned in doc ‘Documentation for ddg_monomer application’ where a helper application ‘minimize_with_cst’ is used.
I have no idea what is the difference between the two approaches and which one is more preferable in practical applications. So anyone can give some details about the backends?
Personally I prefer ‘minimize_with_cst’ because it can keep the original chain IDs in PDB, while ‘relax’ always renumbers the chain IDs to ‘A B C D …’ even the option “-out:file:renumber_pdb false” is used.
Is there any way to let ‘relax’ not to renumber chain IDs?
May 6, 2013 at 7:47 pm #8726Anonymous
Unfortunately, coordinate constrained relax will always renumber PDBs – at least with Rosetta3.4 and before.
The problem is that in order to properly constrain the coordinates of the residues, relax adds and then removes a “virtual root” atom. The process of adding and deleting a residue results caused Rosetta to renumber the entire PDB.
This has been fixed for upcoming Rosetta3.5 release, so coordinate constrained relax will no longer renumber the PDB.
Regarding the difference between constrained relax and minimize, constrained relax is a bit more aggressive in its optimization. It does cycles of packing and minimization with ramping repulsive, so it can better rework hydrogen bonding networks and do a more effective local search. It also consequently takes longer. If speed is not an issue, I’d recommend the constrained relax (though I have access to the version which doesn’t renumber PDBs).
May 12, 2013 at 4:06 am #8750Anonymous
Thank you for the comments.
I am looking forward to version 3.5.
Indeed, constrained relax takes much longer than minimize, and somehow my MPI-compiled version does not boost the calculation. If the MPI works, the speed is more comfortable .
May 12, 2013 at 8:43 pm #8753Anonymous
Most protocols in Rosetta are parallelized at the output structure level – each output structure can be run in parallel, but a single output structure doesn’t gain anything from being run with MPI. There are some protocols that benefit greatly from MPI even for a single output, but relax is one where MPI gains will only be realized when doing multiple output structures.
- You must be logged in to reply to this topic.