Merge pull request #883 from RosettaCommons/kimdn/master_sew
[what has been updated for beta-sandwich_analysis]
* Add a more rule so that two parallel short strands can be "sheet" for Tim-barrel
[what has been updated for SEWING]
* [model generation update]
- sewing::hash_tag_only_terminal_Es is possible
- User can provide an option (generate_five_ss_model) so that 'db -> node (model)' can generate '3_or_5_secondary_structure_based models'
- added unit test (SewingHasherTests.cxxtest) for "db -> node (model/substructure) can generate 5_secondary_structure_based models"
* [Model_trimmer update]
- becomes little faster (don't care rest trimming steps if any preceding trimming decides to erase the model)
- can do 'leave_models_with_E_terminal_ss'
- can do 'leave_parallel_way_H_bonded_models_by_terminal_strands_only'
- can do 'leave_antiparallel_way_H_bonded_models_by_terminal_strands_only'
- can do 'model_should_have_at_least_one_E'
- can do 'model_should_have_at_least_one_E_at_terminal_segment'
* [hashing update] node (model) -> edge (score) can handle 'any number of
secondary_structure' based models
* [model assembly update]
- User can choose between 27 and 125 boxes for neighborhood_lookup
(-sewing:box_length), both scoring and assembly should use same box numbers
* [ExhaustiveAssemblyMover newly added]
- Enumerate every possible assembly exhaustively
- User can choose 'dump_every_model_for_devel_purpose' option
* [MonteCarloAssemblyMover update]
- User can choose not to stick to best_score if specified 'min_assembly_score' as an option in parser
- User can choose 'offset_bump_dsq' as an option
[beautified some other's files]
beautified except
File utility/json_spirit/json_spirit_writer_template.h could not be beautified
File utility/json_spirit/json_spirit_error_position.h could not be beautified
File utility/json_spirit/json_spirit_writer_options.h could not be beautified
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #880 from RosettaCommons/vmullig/reset_issue
Fix the functionality of RESET
The commutativity-breaking resfile command RESET, which is necessary for noncanonical design (since noncanonicals can only be turned ON while canonicals can only be turned off) wasn't resetting canonicals properly. This command now:
- Disables all noncanonicals (as it did before).
- Enables all canonicals (which it was not doing properly before).
- Enables design and repacking.
Which is what the documentation claimed that it was doing.
I've also moved the RESET functionality to a single function in the ResidueLevelTask class, so that we don't have a lot of ways for external things to break commutativity. (Anything can call the reset() function, but it will always break commutativity in exactly this way.)
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #879 from RosettaCommons/vmullig/bintransition_bug
Caught a mistake in BinTransitionCalculator.cc.
This was affecting the setup of sub-bin cumulative probability distributions for the i+1 residue.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #877 from RosettaCommons/rhiju/fix_up_to_virtual_phosphate
Rhiju/fix up to virtual phosphate
changes to integration tests are expected – due to virtualization of 5' phosphates of RNA that should have been happening anyway.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #870 from RosettaCommons/vmullig/peppredict_boinc
Add checkpointing to simple_peptide_predict app
David Kim rightly pointed out that my peptide structure prediction app isn't checkpointing itself properly. While results would still be returned on BOINC, trajectories would likely be repeated if users stopped a work unit and re-started it. This will address that.
Tasks:
- Figure out how the BOINC stuff currently talks to the classic ab initio JobDistributor and to JD2. (I'll probably want to use the nstruct-level checkpointing of JD2, but first I want to understand how everything currently works.)
- Figure out whether JD2's nstruct-level checkpointing somehow knows to skip jobs that failed. (I'll definitely need this, since my app heavily filters, so that many trajectories return no result.)
- Get simple_cycpep_predict to checkpoint. (It was simpler to add checkpointing than to get it to use an existing JobDistributor, and I was able to time the checkpointing to my short jobs rather than the over-frequent checkpointing that I'd have if I relied on the BOINCJobDistributor).
- Get the minirosetta with the simple_cycpep_predict protocol to talk properly to the text overlays in the BOINC graphics.
- Test with the text overlay on the graphics (not tested yet).
- While I'm at it, tweak the energy and RMSD updating in the graphics display.
- Beauty
- Documentation
Cosmetic change to simple_cycpep_predict integration test expected (one extra line of output).
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #862 from RosettaCommons/revert-860-revert-676-tjacobs2/sewing_motifs
Merge of the SEWING framework for de novo protein design. Thanks to Sergey for fixing an issue with the Windows PyRosetta build. All tests pass. Expected changes to features and features_database_schema integrations tests. Also, two new tests for the continuous and discontinuous sewing hasher.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #871 from RosettaCommons/roccomoretti/valgrind_mpi
Fix integration.py w/r/t valgrind and MPI
Not all test directories have non-MPI command files - don't fail if one isn't present
notify author
notify list [rosetta-logs@googlegroups.com]