Merge pull request #1933 from RosettaCommons/sergey/f
Fixing mpi integration test script for cases when more then one extras was specified as platform.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1923 from RosettaCommons/aleaverfay/jd3_silent_file_io
aleaverfay/jd3_silent_file_io
This pull request addresses three significant shortcomings of JD3:
- a lack of silent file support,
- the StandardJobQueen caches stored Poses if they were read from PDBs,
- the StandardJobQueen had no way to deallocate Poses once they were done with
This PR addresses these all at once.
The JobDistributor now allows the JobQueen to send inter-process "deallocation messages," which it will deliver to the JobQueens on remote nodes. These messages will be used by the StandardJobQueen, for instance, to deallocate Poses that are no longer being used. In the near future, the SJQ will also use these messages to deallocate no-longer-needed Resources being held by the ResourceManager.
The JobQueens are now obliged to assign PoseIDs to each of their poses so that they can then use the PoseDeallocationMessage as a way to signal to remote nodes that a Pose is no longer needed. The SJQ does this. Thus she will now deallocate Poses that were read in from either a PDB or a SilentFile. One drawback, however, is that if a Pose is read in multiple times in separate <Job>...</Job> blocks within a job-definition file, then it will be read from disk multiple times. However, it will only be read from disk a single time per process if it appears only once in a job-def file, or if it is input using the command line (e.g. in -in:file:s, or -in:file:silent), regardless of the "nstruct" count.
There is now a SilentFilePoseInputter that delivers Poses to the JQ. Before this PR is closed, there will be a SilentFilePoseOutputter, too. To allow different Pose input options for different jobs within a single job definition file, the structure of silent file reading had to be altered so that the global options system could be wholly avoided if desired. To that end, a new class SilentFileOptions, now can be initialized with a default constructor (to read from the global options system) or from a (possibly-non-global) OptionCollection object. It also has a "parse-my-tag"-like function to read from a TagCOP. AFTER SILENT FILE OPTION CONSTRUCTION, NO PART OF SILENT FILE IO READS DIRECTLY FROM THE OPTIONS SYSTEM.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1932 from RosettaCommons/everyday847/minor_farfar_homology_bug
Fix a minor bug with fragment homology elimination and noncanonicals
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1928 from RosettaCommons/Cyrus/mergetarget
From Steven Lewis:
This is mostly small features I did in in the summer/fall of 2016 and never got around to testing and PRing until just now.
Features:
* smlewis/fix_README - improving the front README, based on a better understanding of our forking policy, thanks to @aleaverfay @kwikwag
* smlewis/FT_clangstatic_fix - fix a clangstatic bug @roccomoretti pointed out (it was a copy paste error)
* smlewis/cyrus/ss_printer_for_commons - fixes to the HELIX/SHEET printer so that the records are ordered properly in large/multichain structures where the numbering restarts per chain. This integration test (secondary_structure_output) will change, as it now has a long enough input to have triggered this bug.
smlewis/cyrus/LoopLengthChange_restype: LoopLengthChange takes a residue type char for what to insert/ Resolves RC issue RosettaCommons/main#1700. Adds integration test.
smlewis/cyrus/actccm_atompair_bounded - AddConstraintsToCurrentConformationMover constraint choice improvements. Adds integration test. This code already had an integration test testing different aspects of the Mover; I made a second test so that it multithreads more efficiently.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1924 from RosettaCommons/rhiju/minor_stepwise_fixes_rna_puzzle18_movie
Rhiju/minor stepwise fixes rna puzzle18 movie
one integration test changes (align_pdb no longer being treated as 'pseudo-native')
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1922 from RosettaCommons/everyday847/immediately_valuable_alterations
Immediately valuable code improvement
Vikram K. Mulligan 8 years The MPI-mode integration test failure is spurious. It's a little problem with the test that I need to fix that's occasionally causing spurious changes, and can safely be ignored.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1920 from RosettaCommons/rfalford12/PyMol_to_PyMOL
Fix Misspellings of PyMOL
=======================
PyMOL is spelled with a capital MOL - however, some recent PyRosetta4 code uses PyMol. We decided that our Rosetta naming should be consistent with the name of the program. So, I renamed every instance of PyMol to PyMOL. These changes will also be reflected in the newest version of the PyRosetta workshops.
Vikram K. Mulligan 8 years The MPI-mode integration test failure is spurious. It's a little problem with the test that I need to fix that's occasionally causing spurious changes, and can safely be ignored.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1918 from RosettaCommons/rhiju/reorganize_sampler_dirs
Rhiju/reorganize sampler dirs
+ cppcheck issues, but stemming from @everyday847 prior commit.
+ some code duplication, but will remove in next PR's.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1913 from RosettaCommons/vmullig/cycpep_align
Add a mover to align a quasi-symmetric peptide to the origin and Z-axis, for true symmetry setup
Currently I do this with some manual editing of PDB files, a script, and the minimizer, but a specialized mover could do this more easily and more efficiently. This will be a needed part of the symmetric cyclic peptide design pipeline.
Tasks:
- [x] Create the mover with Jared's code template scripts.
- [x] Register it.
- [x] Add options to parse_my_tag() for specifying symmetry.
- [x] XSD information for these.
- [x] Add length check and symmetry check (with the CycpepSymmetryFilter) to apply() function.
- [x] Add origin alignment to apply() function.
- [x] Add Z-axis alignment to apply() function.
- [x] Ensure above handles non-protein residues properly (ignores them).
- [x] Allow auto-detection of symmetry.
- [x] In parse_my_tag().
- [x] XSD information.
- [x] In apply() function.
- [x] Add option to delete all but a selected repeat:
- [x] To parse_my_tag().
- [x] XSD information.
- [x] To apply() function.
- [x] Ensure this deletes non-protein residues like TBMB. (Symmetric versions can be added back after symmetry is set up, using the ThreefoldCrosslinkerMover.)
- [x] Document this deletion of non-protein residues.
- [x] Unit tests.
- [x] Integration test.
- [x] Beauty.
- [x] Documentation.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1914 from RosettaCommons/rhiju/restore_thermal_sampler_original
Rhiju/restore thermal sampler original
`thermal_sampler` integration test changes -- actually now it reproduces exactly the behavior of `thermal_sampler_original.cc` (as developed by Fang & Kalli in 2014-2015).
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1910 from RosettaCommons/rhiju/recces_mover
Rhiju/recces mover
Some code duplication introduced (but will eventually delete prior code -- this is part of a bigger refactoring project).
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1846 from RosettaCommons/vmullig/symm_predict
Adding options to simple_cycpep_predict app for only sampling symmetric conformations. Integration test changes to simple_cycpep_predict integration tests are expected.
Tasks:
- [x] Add the options.
- [x] Have them do something.
- [x] Integration test.
- [x] Add mirroring integration test.
- [x] Mute unnecessary output in integration test.
- [x] Find and fix bug in CycpepSymmetryFilter.
- Added a unit test for this, but can't really reproduce the bug, if it exists. It could be that it was a problem with where the filter was applied, and that the asymmetry was introduced in the final relaxation...
- [x] Add GenKIC perturber for copying DoFs (addresses issue #1877).
- [x] Integration test
- [x] Allow perturber to steal DoFs from non-loop residues, too.
- [x] Ensure that anchor_res DoFs are properly grabbed in simple_cycpep_predict.
- [x] Add mirroring option.
- [x] Add mirroring option to integration test.
- [x] Hook this in to the simple_cycpep_predict app in symmetry mode.
- [x] Document GenKIC perturber.
- [x] Document new flags for simple_cycpep_predict.
- [x] Have simple_cycpep_predict ignore linkers when doing RMSD calculation (addresses issue #1879).
- [x] Add option to simple_cycpep_predict to connect ALL cysteines with TBMB (addresses issue #1879).
- [x] Document this.
- [x] Fix issue with output structure numbering in simple_cycpep_predict, so that every output structure in MPI mode has a unique name and number.
- [x] Beauty.
- [x] Documentation.
Put off to a future pull request:
- Add support for multiple natives to simple_cycpep_predict.
notify author
notify list [rosetta-logs@googlegroups.com]