Pyrosetta 2.0 - Dunbrak library binary not loading - strange repacking score

4 posts / 0 new
Pyrosetta 2.0 - Dunbrak library binary not loading - strange repacking score
#1

All,

I am getting large energy increases when repacking a protein using v2.0 of PyRosetta.ScientificLinux64bit (I'm running RHEL5.5_64, Python 2.66)

1st issue On init(), I recieve:

 basic.io.database: Database file opened: .../minirosetta_database/Rama_smooth_dyn.dat_ss_6.4 core.chemical.ResidueTypeSet: Finished initializing fa_standard residue type set. Created 7533 residue types core.pack.task: Packer task: initialize from command line() Warning: Unable to locate database file Dunbrack02.lib.bin core.pack.dunbrack: Dunbrack library took 1.96 seconds to load from ASCII Warning: Unable to locate database file Dunbrack02.lib.bin Warning: Unable to locate database file Dunbrack02.lib.bin core.pack.dunbrack: Opening file .../minirosetta_database/dun02twlyHM for output. core.pack.dunbrack: Moving temporary file to .../minirosetta_database/Dunbrack02.lib.bin core.pack.task: Packer task: initialize from command line()

Which, according to a previous post, could be from a file permissions issue, although everything seems fine and PyRosetta is in my home directory.

2nd issue Here is a snippet from my repacking:

 scorefxn = create_score_function_ws_patch('standard','score12') task_pack = standard_packer_task(pose) task_pack.restrict_to_repacking() packmover = PackRotamersMover(scorefxn,task_pack) packmover.apply(pose) pscore = scorefxn(pose)

Repacking the structure causes a 50% positive increase in score. Using PyRosetta v1.0, I do not have this issue.

Can I try copying over the v1.0 Dunbrak binary?

Thanks,
Mike

Post Situation:
Wed, 2011-05-11 06:24
msellers

A) You can remove the dunbrack binary at any time; Rosetta should automatically regenerate it. The binary contains no calculation; it just exists to speed loading (usually 30 seconds for text, versus 0.1 for binary).

B) The file permission is at issue if it ALWAYS has to load from text. If the binary exists, I don't think it's the issue.

C) You should try using the PackerTask option for input sidechains. It's use_input_sc or include_current or something similar. That will leave the native sidechains as allowed rotamers; if your score does NOT increase it may mean that you have many off-rotamer sidechains in your input that Rosetta can't find when packing.

Wed, 2011-05-11 09:05
smlewis

I will try the "leave native sidechains"...

What's funny about this is that PyRosettav1.1 (Rosetta r34968) does not have this problem. The poses I'm working with are two docked chains that are separated into differend PDBs and repacked to compute a binding energy. Docking was run with PyRosetta v1.1, so perhaps something has changed in the Dunbrack library?

Wed, 2011-05-11 10:24
msellers

I don't think we've made any changes to the Dunbrack library in between PyRosetta versions.

The Dunbrack08 library was an addition to the Rosetta3 source internally at some point after we got serious about using Rosetta3. That's the only change I know of. You (and most people) are using Dunbrack02, which is shared back with Rosetta++ and hasn't been changed.

Wed, 2011-05-11 11:15
smlewis