Minor changes to the cartbonded potential.
- Adding in a backbone O-C-N-H torsional potential.
- Many of the planar restraints are now period 1 instead of period 2. Previously, badly clashing structures, when minimized, would cause these groups to flip. This will no longer happen.
- Greatly reducing some of the hydrogen torsional potential weights. Doing so sees modest improvements in non-ideal rotamer recovery.
- Uniformly increasing bond angle weights. This shows a similar improvement in nonideal RR.
Updating the unit test to reflect these scorefunction changes.
All nonideal integration tests will show changes.
Thanks to Hahnbeom who did most of the work for this commit.
notify author
notify list [rosetta-logs@googlegroups.com]
Removing unused flags from oop_create integration test
Removed mm_std weights and resfile from commandline flags after cleanup
of options system. The score function in oop_create is hard coded to
mm_std so the commandline weights flag was being ignored. The resfile
can be used during a final repack in oop_create but not in this
integration test.
oop_create integration test will change in the log file because the
options warning will no longer print. Also, the commentted out options
in the flags file and deletion of res file and unused constraints file
will show changes. The remainder of the log file should be the same
though.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #65 from RosettaCommons/roccomoretti/option_cleanup
Option System Cleanup
Reorganize options_rosetta.py to be a bit more ... organized. It's now ordered: major option groups (in, out, run, jd2, score, packing) first, then other general option groups in alphabetical order, then application-specific option groups in alphabetical order, and then personal/pilot option groups in alphabetical order. (Classification was done by best guess, and is not an official pronouncement. Feel free to reorganize. Also, alphabetical ordering is not absolute - logical regroupings can be permitted.) Some formatting cleanup (tabs, not spaces, despite being a .py file) was also done.
Additionally, approximately 200 options which were never referenced on the C++ level were commented out. Use of these options will now result in an error when you try to run a protocol with them. (Thanks to Luki for help with the clang AST parser, which was used to ferret out unreferenced options.)
Finally, a new option "-run:show_unused_options" was added. When set, at the end of the run a list of options which were user set but never actually referenced by the program will be printed out. The flag is unneeded for JD2 protocols, which will automatically print out the list of user-set-but-unused options at the end of a successful run with at least one output structure. (Thanks to Doug for suggesting this feature.)
No unit test changes, but a large number of integration test changes. These are due either to removed options being removed from the flag file for integration tests, or from the new default-on-in-jd2 behavior of the show_unused_options feature, which touched a large number of tests. (A fair number of non-JD2 integration tests also have user-set-but-unused-options, but are silent in this change.)
notify author
notify list [rosetta-logs@googlegroups.com]
Updated the follwoing integration tests:
1. splice_out_generic
2. splice_out
3. splice_in_genric
4. splice_in
5. splice_seq_const
These ofcourse will fail compared to previous commits
In Splice.cc:
1.Add Carbonyl coordiante constraints
2. Changed the current segment behvaiour so it reads the current segment from the XML rather than the pdb
3. In SpliceSegments the occurrence data for H3 loops with no occurrence data is filled in by Rosetta (previously this caused a segmentation fault)
4. In SequenceProfile I added getter and setter functions for accessing the occurrence_data array
5. In SeqprofCons I added tracer lines to check correctness of occurrence_data array
notify author
notify list [rosetta-logs@googlegroups.com]
Making inv_kin_lig_loop_design add the appropriate cutpoint variants for
compatibility with the CCD mover. This was causing an exception to be
thrown, which in turn caused the inv_kin_lig_loop_design integration
test to fail.
This fixes the inv_kin_lig_loop_design integration test.
notify author
notify list [rosetta-logs@googlegroups.com]
Patch System: added 3 new PatchOperations
NEWS: This push introduces three new PatchOperations to the patch system:
- ADD_ATOM_ALIAS -- allows one to specify alternative atom names that Rosetta
will understand when reading in a PDB file
- ADD_BOND_TYPE -- an alternative to ADD_BOND that allows specification of
bond type
- SET_FORMAL_CHARGE -- sets the formal charge of a specified atom
All of the above new operations rely on code written by R. Moretti and co.; I
simply added the functionality to the patch system as well. (Previously, atom
aliases, bond types, and formal charges could only be added/set from .params
files.)
The above operations were added to carbohydrate patch files were applicable.
The carbohydrates integration test was expanded to specifically test that the
atom aliases work: a pdb with non-Rosetta atom names correctly loads in.
In addition, centroid-mode sugars were updated, (which at this point are
simply copies of the fa_standard ones).
----------------------------------------
Unit test status: Pass
Integration test status: Not Run (Only sugar code makes use of the additions.)
carbohydrates: test expanded
notify author
notify list [rosetta-logs@googlegroups.com]
Loops: bug fix for when overlapping loop is found
In the case when one adds an overlapping Loop to a Loops object, there was a
dormant bug that our recent CCD refactor uncovered where the Loops object
would attempt to output the Loop information after deleting the no-longer
needed loop. I just fixed that.
This push also includes the removal of two deprecated/legacy antibody
integration tests.
notify author
notify list [rosetta-logs@googlegroups.com]