Merge pull request #650 from RosettaCommons/vmullig/repeat_energy
Create an energy term that penalizes repeat stretches (e.g. poly-Q sequences)
Yes, I know this is inherently non-pairwise decomposable. It won't work with the packer in its present form, but can be used for post-hoc scoring.
Tasks:
- Create forward declarations.
- Create creator.
- Create .cc and .hh files.
- Register these.
- Add energy type to the scoretype enum.
- Write compute function.
- With hooks for rapid computation given a vector of ResidueCOPs.
- Flag for config file (with default version in database) that determines how the penalty is applied.
- Default config file in database.
- Unit tests.
- Documentation.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #643 from RosettaCommons/BobbyLangan/sasa_bug_fix_pdbinfo
Fixes pdb_info issue in SelectBySASA operation
Added a pointer check before trying to fetch pdb_info in SASA operation
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #649 from RosettaCommons/vmullig/mpi_c_bool_issue
Switching all instances of MPI_C_BOOL to MPI_CHAR
Switching all instances of MPI_C_BOOL to MPI_CHAR to fix compilation issues on certain systems. I'm using MPI_CHAR for my dummy/acknowledgement signal because a char is the one datatype that's guaranteed to be one byte wide in the C++ standard, and I don't want to waste time sending more than one byte for an acknowledgement signal. (Unfortunately, the C standard did NOT originally have a boolean data type, so MPI_C_BOOL is not always defined in all flavours of MPI, apparently. Still, a C++ bool is 8 bits wide rather than 1 bit, since the smallest memory space that one can address is a byte, and a bool* wouldn't make sense if bools were 1 bit. This means that I'm not really losing anything by switchng this to a char.)
No test changes expected.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #635 from RosettaCommons/vmullig/new_ncaas
Adding a few more beta- and gamma-amino acids to the database
Mostly cyclics (e.g. trans-ACPC).
...And in the process I caught a bug in the helical bundle code. Fixed. This causes the multirepeat integration tests to fail.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #642 from RosettaCommons/vmullig/zibo_bundlegridfilter
Create a filter for reporting helical bundle parameters.
This creates a new filter, the BundleReporterFilter. The main purpose of this filter is to look at a parametric pose and dump out (a) the bundle parameters, and (b) the Rosetta energy, given a scorefunction. By default, the filter always returns true, though it can be set to return false always, or to filter based on the energy (much like the ScoreType filter).
Tasks:
Create the filter files (-.cc, -.hh, -.fwd.hh, -Creator.hh)
Register them with the appropriate init files.
Get the filter to write out parameters.
Handle the case of the filter receiving a non-parametric pose.
Get the filter to score and write out the score.
Get the filter to filter on score, if the user so wishes.
Integration test.
Documentation.
Also:
Fixed problems with the ScoreType filter (uninitialized variables).
notify author
notify list [rosetta-logs@googlegroups.com]
Fixing a stupid hbs_dock_design bug (somehow I wasn't initializing the Mover; I'm not sure how this ever worked--maybe because I mostly invoke it using NcbbDockDesignProtocol now) and adding an AtomType found in the HBS linker to the split_unfolded default energies
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #628 from RosettaCommons/kylebarlow/general_dev
Fixing -Wpotentially-evaluated-expression warnings-as-errors added in Clang-3.6
Some of these occurred in the cppdb external code. I first tried updating this code to the newest released version, but this didn't fix the warnings, and also broke a pilot app, so I reverted that change.
I suppressed the warnings in the external code, as we do not maintain it, and it doesn't look like a warning that really matters anyways.
In the Rosetta code, I think the warning was being caused by a deference of an owning pointer not playing well with typeid(), so I fixed the warning by first creating a temporary reference to typeid(). We seem to be good OOPers for the most part, so there aren't many uses of typeid() in the codebase to bring this up again.
notify author
notify list [rosetta-logs@googlegroups.com]
FoldTree: Refactor Part II
NEWS: FoldTree no longer outputs with its own endl
In addition, FoldTree unit tests have begun to be expanded.
You can monitor the ongoing progress of this refactor at pull request #630.
----------------------------------------
Unit test status: Pass
Integration test status: ~80 changed tests
notify author
notify list [rosetta-logs@googlegroups.com]