「view this page in B3 βῆτα server」

Revisions №58468

branch: master 「№58468」
Commited by: Jared Adolf-Bryfogle
GitHub commit link: 「763a01403a7a6aa8」 「№1045」
Difference from previous tested commit:  code diff
Commit date: 2016-02-18 14:26:27

Merge pull request #1045 from RosettaCommons/raemisch/fix_remodel Remodel Fixes A few fixes for various things in remodel from Sebastian Raemisch (@raemisch) and I. Heres the list: 1) Remodel should work on the first chain of the pose or the chain given in -run:chain. However, (in some instances) you must pass run:chain even if you only have a single chain in the PDB. This fixes that, and by default will use the first chain of the PDB. 2) If you use Alanine's instead of any other aa for centroid-stage backbone fragment building using the -remodel::generic_aa option, you would seqfault. This fixes that. 3) Lets say you have an option, _design_neighbors_, that you explicitly set to false on the command-line. Remodel would design those neighbor residues and baffle you. The bug is that for boolean options all over remodel code, instead of accessing the option itself, it accessed the use of the option. This fixes it in every file that used the remodel options. In most instances (as these were default false or were intended to be default false), this is OK. However, this is an interface bug, and one should generally __NEVER__ do this as it can (easily) lead to wrong results. Remodel integration test will change - It was tested as is (in the PR) and then updated to use -out:prefix in order to run its 3 tests and have good comparisons for the future.

...