Merge pull request #1245 from RosettaCommons/JWLabonte/sugars/general
PyRosetta/Carbohydrates: Adding sample files/tests for sugars to ship with PyRosetta
This merge adds a new directory to PyRosetta and a new unit test for carbohydrate functionality within PyRosetta. It also includes a miscellany of other sugar- and/or PyRosetta-specific tweaks/improvements/corrections.
This merge also fixes a bug in the Windows build, because apparently "rad2" is a keyword for Windows! Thanks to @lyskov
All unit tests pass. carbohydrates test changes expected.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1252 from RosettaCommons/jadolfbr/sugars/glycan_clash_check
Adding a pilot app specifically for glycan clashes which quantifies clashes between specified glycan branches and other chains and glycan branch - glycan branch clashes. Does not count clashes for intra glycan branch. Does not echo input files, adds info to output scorefile to allow for MPI runs.
Also adding description of Rosetta LJ in AtomType from a description by @aleaverfay in a dev list discussion some years ago (sigma1 vs sigma2) to help future travelers.
Clash Definition
============
Atom is clashing if it clashes with ANY other atom
Count up clashes (hard and soft) per-residue
- Hard:
- distance <= LJ1 + LJ2
- Soft:
- distance <= (LJ1 + LJ2) * (1 - S). S is between 0 and 1. Here, we use .33 as default.
-> Intuitive definition, easily conceptualized, still quantitative
Relevant Options:
======
-glycan_branches
- Required. Vector of glycan branches. Rosetta Residue numbering or PDB like RosettaScripts
-chech_chains
- Required. A list of chains to compute chain-based clashes. If not given, will only compute glycan-glycan clashes!
-ignore_hydrogens
- Should we calculate only heavy-heavy atom clashes?
-soft_clash
- When we calculate atom-atom distances using LJ distances, clash if distance < (atomI_LJ + atomJ_LJ)*(1 - soft_clash)
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1274 from RosettaCommons/JWLabonte/sugars/CAPRI_redux
Carbohydrates: CAPRI Redux Part II
missing file; unit tests pass; carbohydrates test changes expected
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1224 from RosettaCommons/vmullig/binselector
Add a ResidueSelector to select residues based on their torsion bin (e.g. ABEGO bin)
This should be useful for peptide design, as well as for canonical design in which one might want to design positions in the A bin with helix-favouring residues and positions in the B bin with strand-favouring residues, etc.
Tasks:
- Create ResidueSelector with Jared's nifty scripts.
- Add BinSelectorCreator to ResidueSelectorCreators.hh.
- Add ResidueSelector to core.3.src.settings.
- Loading of bin definitions.
- Write apply() function.
- Write parse_my_tag() function.
- Unit tests.
- Selector XSD.
- Integration test.
- Add bin_params files for selecting residues in L- or D-proline bins.
- Test for this, too.
- Option to only select alpha-amino acids.
- Documentation.
- Beauty.
Also:
X Fix PhiSelector bug identified by Parisa --> put off to a future pull request.
- Fix a little bug in the RESET command in ResidueLevelTask_ (unnecessary debug_asserts that tripped things up in debug mode).
- Integration test for PhiSelector, too.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1268 from RosettaCommons/roccomoretti/icc_pseudofix
Change RotamericSingleResidueDunbrackLibrary to avoid unnecessary intermediate data structure.
This should fix the current issue with fa_dun score differences between GCC and ICC.
This doesn't, however, necessarily address the underlying issue of why ICC was breaking the code when optimizing. I've (separately) put together a test case to pass along to Intel as a possible bug report.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1269 from RosettaCommons/jadolfbr/antibodies/skip_cis_trans_cluster_matching
Jadolfbr/antibodies/skip cis trans cluster matching
Adds an option to Rosetta Antibody for North/Dunbrack Clusters:
-allow_omega_mismatches_for_north_clusters
- Skip first grouping by cis and trans designation for clusters in which a cis/trans designation currently does not exist.
This is due to the North clusters being first grouped by the omega cis/trans for each residue, and then clusters within that (for example TTTTTTTCTT ). This directly relates to the sequence profiles used for design. We may eventually make this option default on, but for now it stays off. If a cluster for a CDR cannot be found because of this mismatch, we give a nice warning to users in Red and let them know of this new option.
Also add an integration test for identify_cdr_clusters.
Thanks to @lyskov for helping with the tracer colors. I added some info to the Tracer header that gives examples on how to do it for most common purposes.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1259 from RosettaCommons/JWLabonte/sugars/branching
Post-ChemicalXRW: removing the -read_pdb_link_records option
LINK records are now read always.
All unit tests pass.
Integration test changes expected where options have been changed.
Note that the C-terminal_conjugation patch causes a bug in some unit tests, so it has been commented out of patches.txt until the bug has been squashed. In the meantime, it can be included by the option system.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1262 from RosettaCommons/jadolfbr/antibodies/de_loop_design
Antibody DE Loop Design
Description
========
This PR finally enables Proto CDR4, also known as the DE loop, to sequence design within RosettaAntibodyDesign. L4 and H4 are specified the same way as others within the cdr instruction file. L4 and H4 can also be specified as neighbor CDRs.
'DE' can used as in 'ALL' keyword in the instructions file to set options for both loops. Proto CDR4 is not designed by default.
Testing
=====
Unit tests have been updated to include and test this new functionality, and an integration test has been added. The current antibody_designer integration test has also been shorted by changing the min type.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1264 from RosettaCommons/jadolfbr/sugars/glycan_relax_random_start
Glycan Relax updates
Description
=========
This PR makes the default number of Glycan Relax rounds to 50 and adds two ways to start the modeling off, either by randomly sampling every glycan torsion first from the move map, or by randomly sampling each torsion using the sugarBB probabilities for those with which we have sugarBB data.
@raemisch Has run preliminary tests on 50k models per experiment and found that if you do not have a real starting glycan conformation, that 50 or 75 rounds and random start both produces greater sampling diversity in the final models (with a lower number of decoys needed) and enriches for low-energy models. Still much benchmarking to do, but this is a start.
Options
======
cmd-line:
- -glycan_relax_random_start
- -glycan_relax_sugar_bb_start
RS:
- random_start
- sugar_bb_start
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1263 from RosettaCommons/tlinsky/disulfidize_mutate_to_ala
Disulfidize no longer mutates disulfides outside of specified subsets to ALA
notify author
notify list [rosetta-logs@googlegroups.com]