Merge pull request #769 from RosettaCommons/vmullig/muteall_issue
Fixing B3V.params.
Somehow, I managed to delete a line in B3V.params in my previous commit. Weeeird.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #745 from RosettaCommons/rhiju/remove_RNA_residue_type_set
Getting rid of RNA residue type sets
Last full set of tests: http://benchmark.graylab.jhu.edu/revision?id=2377&branch=
(ran a few more after that, but cancelled after unit & integration. They were just to check merge conflicts.)
Tests look good:
+ Cosmetic change: removal of line loading in rna_residue_type set. But no further changes, as desired.
+ Cosmetic changes: removed unnecessary packer initialization when pack_missing_sidechains() is called but there is nothing to pack. (In a couple tests, exposes warnings about unused options like -packing:use_input_sc.) Sorry, this changes one line in a lot of tests, but I figure folks wouldn't mind removing this extra unnecessary step.
+ cstfile_to_theozyme_pdb, match_1n9l is unstable in integration, unfortunately, despite my prior attempt to fix. Those still run stably in integration.release_debug. Will be setting up a separate branch to try to resolve remaining instabilities.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #763 from RosettaCommons/vmullig/hydrophobic
Adding a HYDROPHOBIC property to the properties table.
This is needed because some residues that are not polar (e.g. glycine, alanine, proline, cysteine) are not really hydrophobic (insofar as they're not major contributors to a well-packed hydrophobic core), either.
POLAR: DEHKNQRST
HYDROPHOBIC: FMILYVW
Neither POLAR nor HYDROPHOBIC (for our purposes): ACGP
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #718 from RosettaCommons/JWLabonte/sugars/CAPRI_redux
CAPRI Round 27 Target 57 Redux
This merge introduces the `.params`, patch, and apps files required to hopefully recapitulate and improve on our CAPRI Round 27 Target 57 results using the new carbohydrates framework.
All unit tests pass.
carbohydrate test changes expected; all other integration tests failing on the test server were either unstable or broken prior to my changes.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #764 from RosettaCommons/vmullig/fix_compile
Fixing the build.
Sergey Ovchinnikov just discovered that the existing build is actually a broken build, despite passing all build tests on the test server and on my machine. The problem is that I removed the sole member of an option group, so that that option file is no longer generated by the options.py script. There are a few files that include it, though. The file is there if it has ever been generated (e.g. in all test server builds, and on any existing checkout of Rosetta), so Rosetta compiles happily, but it is missing on new checkouts, so compilation fails there.
I test-built this after deleting all of the auto-generated options files and regenerating them. Will merge without testing, to get master building properly as quickly as possible.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #757 from RosettaCommons/vmullig/fix_invrotremodel
Fix the inverse_rotamer_remodel integration test
This seems to be failing due to a bunch of missing mm_std bond angle parameters. It bothers me, though, that this was NOT failing on the Linux builds. There was an option for ignoring missing bond angle parameters, but (a) this was set to false, and (b) even if this were set to true, you should end up with a segfault on any system if you ignore the missing parameters. Instead, the segfault was only happening on Mac. I don't know why missing bond angle parameters were being ignored, or why the segfault was only on Mac.
I've removed the option (so there should ALWAYS be an error thrown and an abort if there are missing bond angle parameters), and have added the missing bond angle parameters to the database (copying similar bond angle parameters and adding comments indicating that these are copies).
notify author
notify list [rosetta-logs@googlegroups.com]
Fix cppcheck tests
Delete cached cppcheck output files if the corresponding source file
is no longer present. This fixes issues in the commits branch on the
testing server where spurious cppcheck errors are being reported.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #708 from RosettaCommons/jkleman/mp_updates2
Jkleman/mp updates2
= protocols/membrane/util:
added functions pose_tm_com, rsd_closest_to_pose_tm_com, create_membrane_foldtree_anchor_pose_tmcom
added unit tests for them
= MembranePositionFromTopologyMover:
cleaned up foldtree code and created new foldtree
likely to cause integration test changes in mpfastrelax and depending protocols
= OptimizeMembranePositionMover:
cleaned up foldtree code and created new foldtree
likely to cause integration test changes in depending protocols
transformed pose into membrane first instead of computing membrane from topology, this is much more correct because the center and normal search depends on it
= TransformIntoMembraneMover:
removed the option of optimizing the embedding within the mover, this is much cleaner and doesn't lead to circular calling of OptimizeMembraneMover and TransformIntoMembraneMover
= TranslationRotationMover:
added checks
= integration tests:
added normalize_to_thk flag to integration tests!
added integration test for mp_interface_statistics
added integration test for mp_transform_optimize
added integration test for visualizing embedding
Apps
= mp_transform:
debugged version where the membrane position is optimized in addition to just transforming it into the default membrane this is now not in JD2 any more because the protocol differs and I need to access MEM after optimizing the membrane this also depended on leaving optimization and transformation more or less independent
= helix_from_sequence:
removed option to optimize embedding, was incorrect and will be added back soon
integration test changes: mostly cosmetic, otherwise expected
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #758 from RosettaCommons/rfalford12/make_meminfo_default_constructor_private
Making the default membrane info constructor private
The MembraneInfo object relies on data initialized in the AddMembraneMover#apply() function. For this reason, it doesn't make sense to create a default MembraneInfo() object so I'm making the constructor private.
notify author
notify list [rosetta-logs@googlegroups.com]