Bugfix for RosettaCM and branched glycans (#221)
RosettaCM, with a branched glycan ligand, was segfaulting. In 2 locations, the "previous residue" of the given polymer entity was selected by choosing the position prior to the one in question, which is not appropriate for a branched glycan. The fix was to set a conditional for carbohydrates that uses conformation::carbohydrates::find_seqpos_of_saccharides_parent_residue to find the previous residue.
notify author
notify list [rosetta-logs@googlegroups.com]
Attempt to fix failing tests in main. (#299)
This PR attempts to address (in part) the following failures:
* code_template_tests_src integration -- Update test (and script documentation) for the new scheme
* unit valgrind -- fix for Python3
* clang tidy -- update NULL/nullptr
* mp_f19_tilt_angle scientific -- Make more robust, and remove unneeded (for the scientific test) repacking step added with RosettaCommons/main#5551
* protein_data_bank_diagnostic scientific -- Errors are due to running out of pseudotty handles, attempt to recognize that situation and use a fall-back approach.
* glycan_structure_prediction scientific -- Adjust number of processors and runtime to run on single node.
I've also made some changes to better support the GCC 11 & Rocky9 build I'm now doing locally.
I've also attempted to fix the replica_docking MPI integration test timeout, but there some additional issues which mean it's not quite working correctly.
notify author
notify list [rosetta-logs@googlegroups.com]
removing "defaults" Conda channel in favor of "conda-forge" (#241)
Conda recently changed the licensing terms for they "defaults" channel so let use conda-forge instead.
notify author
notify list [rosetta-logs@googlegroups.com]
Fix issue related to PackerTask usage in homodimer_design. (#115)
With a -pack_min_runs of 2 or greater, there's a chance that the sequence of the structure changes. This is a problem for a fixed PackerTask run, as `task_is_valid( pose )` fails if the base residue types don't match up properly. This is an issue found in Discussion #112
The fix is simple -- just use the TaskFactory which we already have lying around instead of manually generating the packer task. (The PackRotamersMover will properly re-apply the TaskFactory.) We can also make the interface of PackRotamersMover a bit nicer.
notify author
notify list [rosetta-logs@googlegroups.com]