Merge pull request #2586 from RosettaCommons/jadolfbr/fix_jd3
Fix MPI-JD3 overwrite behavior
The MPIWorkPoolJD was not properly checkpointing in order to continue the run, and overwriting already-complete files. This fix is from @aleaverfay.
Thanks Andrew!
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #2556 from RosettaCommons/roccomoretti/xml_to_MMfactory
Change parse_my_tag() to use MoveMapFactories.
The parse_movemap() functionality used in parse_my_tag() took a Pose. This was less than desirable, as there's no guarantee that the pose that the move map will be used on will be related in any significant way to the input pose being passed to parse_my_tag().
Now the parse_my_tag() will generate a core::select::movemap::MoveMapFactory object from the tag specification, rather than a MoveMap itself. This will be then applied against the pose in the apply() function to get a MoveMap that is actually used. By waiting for apply() (and re-generating the MoveMap for each call to apply), the MoveMap thus generated will reflect the status of the pose as it is then, rather than the pose as it was in the input.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #2576 from RosettaCommons/Cyrus/mergetarget
Cyrus/mergetarget
From Steven Lewis:
* minor tweaks to unit test code template
* interface tweaks to CoupledMovesProtocol - allow setting of scorefunction/taskfactory externally. Note that the ctor of CoupledMovesProtocol does custom setup on the scorefunction, presumably to support the angle dependencies of Backrub. This was refactored into a standalone configure function. parse_my_tag is routed through this to address https://github.com/RosettaCommons/main/issues/2455 @anumazam
* FillAUTOTaskOperation. CoupledMovesProtocol uses ClashBasedRepackShellSelector to automatically select what residues should repack; the functionality only works when using a resfile with the AUTO command. The TaskOperation allows you to activate this without an actual resfile by filling in the AUTO command’s effect. (This is the most conservative way to fix this problem, as requested by others interested in this code; in lieu of https://github.com/RosettaCommons/main/issues/2434). Note a change to line 247 of CoupledMovesProtocol (sorry, no link, I presume it would break) to support this. Unit test for FillAUTOTO.
* bugfix for https://github.com/RosettaCommons/main/issues/2494. Unit test for LoopAnalyzerMover to prevent recurrence. I had to change the reference values in the unit tests after updating from master; I think they were affected by https://github.com/RosettaCommons/main/pull/2309 (the NYC hackathon). Discovered an insufficiency in Loops:verify_against in writing the utest; improved its range of problem detection.
Test implications:
unit tests: two new ones
integration tests:
expected changes due to LoopAnalyzerMover (gains extra output)
FAIL LoopAnalyzer
FAIL AnchoredDesign
FAIL AnchoredPDBCreator
FAIL loop_creation
These need to be watched on the test server but probably won’t change:
failing in on my test machine due to some weird binary file; probably spurious:
FAIL backrub_interface_ddG
FAIL InterfaceDdG
FAIL cluster_calibur
fails on my test machine because of its git dependency:
FAIL code_template_tests_unit
FAIL code_template_tests_app
FAIL code_template_tests_src
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #2560 from RosettaCommons/smlewis/cyrus/XMLs_from_XRW_rosettascriptstestbranch
Update rosetta_scripts_scripts submodule with cyrus-sourced XSD XRW scripts. Also update with a bunch of other people's scripts, apparently. I'm tired of the "wait for test server results, oh, someone has changed the base version, you have to test again" dance.
notify author
notify list [rosetta-logs@googlegroups.com]