Merge pull request #1988 from RosettaCommons/dimaio/cart_bonded_threaded
Four real integration test changes. All verified to be due to change to cart_bonded_improper potential.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1990 from RosettaCommons/roccomoretti/fix_AntibodyDatabaseManager
AntibodyDatabaseManager: extend temporaries until after we're done with the query.
In cppdb::statement::bind(), the passed value should not be a temporary return value from a function. If it is, it will be destroyed before the query completes, leading to havoc and bad results. You need to extend the lifetime of the returned value by assigning it to a variable so it doesn't get destroyed before you're done with the query.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1982 from RosettaCommons/roccomoretti/ubsan_fixes
Fix two issues which were causing errors with the undefined behavior checker.
In LoopProtocol, explicitly cast the regex match to a string, rather than whatever odd type it is under the hood.
In MotifGraftMover, explicitly initialize variables in a plain-data struct, as the uninitialized values were causing issues when the structures were copied. ("2" is not a bool.)
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1985 from RosettaCommons/aleaverfay/cleanup_nblist_cutoff_logic
Refactoring neighborlist construction logic in EtableEnergy
This PR will assuredly cause numerous intgration-test trajectory changes, as the
cutoff parameter controlling hydrogen motion tolerance is increased from its previous
0.25A to 0.75 to match the parameters for heavy-atom motion. This means that fewer
interactions involving hydrogen are going to be ignored during minimization; i.e.
the scores and derivatives will change and thus trajectories will change.
This commit is needed by PR #1972, but because of its dramatic changes to integration
tests, I wanted to commit it by itself.
Vikram K. Mulligan 8 years Looks like this broke three splice_out integration tests (i.e. prevented them from running).
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1978 from RosettaCommons/vmullig/tweak_peppred
Adding a final symmetry filtering step to simple_cycpep_predict
When sampling with quasi-symmetry, sometimes the final relaxation step would break symmetry. This adds one more filtration step based on symmetry after that relaxation, so that asymmetric poses are thrown out.
Tasks:
- Add extra filtering step.
- Double-check that BOINC build is still happy.
- I think it is, but I have an older version of the Linux BOINC libraries built. Will check on Ralph the next time that Rosetta is deployed.
- Also, add filtering to bettercluster.cc.
- Add threshold option to bettercluster.cc.
- Beauty.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1971 from RosettaCommons/jadolfbr/fix_append_pose_by_jump
This fixes a bug where energies were not being correctly cleared when appending the pose by a jump. Most other append/delete functions are calling this clear properly.
This led to an issue where the concatenated pose was not able to be dumped properly.
Found by @lqtza and confirmed as missing in the function by @roccomoretti
The two functions where we now properly clear the energies are:
- append_pose_by_jump
- delete_residue_range_slow
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1976 from RosettaCommons/roccomoretti/sicdock_uninitialized
Fix 'may be used uninitialized' compiler error
These were causing issues with gcc 5.4
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1970 from RosettaCommons/roccomoretti/pertmin
Roberto Chica inspired random perturbation followed by minimization mover for RosettaScripts.
Potentially useful for exploring local conformational space.
1. Davey & Chica, Proteins 82:771-784 doi:10.1002/prot.24457
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1943 from RosettaCommons/aleaverfay/jd3_derived_jq_specifies_inputter_subset
Rebecca asked for a JD3 feature that would allow her membrane-related application to require a particular PoseInputter be used so that it would be able to initialize the Pose in a special way to ensure that all of the membrane-related data were in place at startup. This pull request expands the set of features that the StandardJobQueen (SJQ) provides for classes wishing to derive from her.
The derived JobQueen (DJQ) will call one or both of two functions in her constructor to inform the SJQ base class of what pose inputters she would like to allow, and will call one, two, or three functions to inform the SJQ base class what pose outputters she would like to allow.
It is possible to for the DJQ to say:
"I would like all of the inputters registered with the PoseInputterFactory",
"I would like all of the inputters registered with the factory, and these X, Y, and Z additional inputters that are not registered", or
"I would like this list of inputters which may or may not already be registered with the factory"
@rfalford12
notify author
notify list [rosetta-logs@googlegroups.com]