Merge pull request #2181 from RosettaCommons/weitzner/more-py3
Weitzner/more py3 -- a beautification test (related to LK Ball) and one integration test (ligand_database_io ) failed on this PR, but I only touched some python code related to testing, so I'm merging this thing in.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #2125 from RosettaCommons/dimaio/more_beta_nov16_bugfixes
This pull request fixes several minor bugs with beta_nov16 (mostly lk_ball_bridge):
1) The ramp widths for lk_ball_bridge and lk_ball (which are different) were used incorrectly in places, ultimately leading to derivative discontinuities in lk_ball_bridge. Also modifying ramp widths to best match the shape of lk_ball_bridge that was previously used.
2) The heavyatom-heavyatom interaction radius for lk_ball_bridge is a tiny bit larger than that of fa_sol; reusing fa_sol's minimizer graph missed some interactions. Modifying to have lk_ball create its own minimizer graph.
3) countpair was not correctly applied for lk_ball_bridge_uncpl in scoring (but was in packing/minimization) ... this has also been corrected. Also fixed a small countpair inconsistency with lk_ball across non-polymer connections.
This change only effects beta_nov16 scoring with one exception: beta_nov15 will report slightly different scores in disulfide-containing structures. This leads to several beta_nov15 integration test failures: disulfidize_beta_cys, genkic_ramaprepro_sampling, simple_cycpep_predict_tbmb, ppi_*
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #2168 from RosettaCommons/aleaverfay/rotsets_offsets_bug
bugfix w/ deleting and adding rotamers
There's a bug in the way the RotamerSets class handles the addition
and deletion of Rotamers from the RotamerSet objects it holds. This fix here
cause the RotamerSets object to make sure its rotamer-count and offset data
is up to date before setup_for_packing is called, (and therefore, before
simulated annealing begins).
I had been really hoping this bugfix would have turned up more integration test changes, as I have another PR in the works (#2165) that produces quite a few trajectory changes, and I'm not yet certain why.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #2169 from RosettaCommons/vmullig/fix_zerolength_bug
Fixing an error in Conformation::rebuild_polymer_bond_dependent_atoms
Fixing an error in Conformation::rebuild_polymer_bond_dependent_atoms() in the edge case that this function is called on an open terminus. (This was creating problems with using the DeclareBond mover to create terminal disulfide bonds).
@csykang : This should fix the problem with your script, and allow you to build disulfide-cyclized peptides without extra tail glycine residues.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1508 from RosettaCommons/vmullig/support_nmethyl
Adding support for N-methyl amino acids
# **Background and Rationale:**
N-methyl amino acids can be used during peptide synthesis to confer various desirable properties on a synthetic peptide. N-methylation removes a hydrogen bond donor, in many cases promoting solubility in hydrophobic environments and helping with membrane permeability. It also greatly alters the conformational preferences of an amino acid residue.
We want to be able to design with N-methyl amino acids, and this pull request is intended to add support for:
- Modelling N-methyl amino acid geometry (adding a patch for N-methylation).
- Properly scoring N-methyl amino acid-containing peptides (permitting the loading of custom Ramachandran and p_aa_pp tables for N-methyl amino acids).
Following the implementation of PackerPalettes (pull request #1047), a future pull request will add:
- Support for designing with N-methyl amino acids (_i.e._ getting the packer decide whether to N-methylate a position or not).
**Tasks (note that many of these changes have already been merged into master in incremental merges):**
- [x] Add a N-methylation VariantType.
- [x] Add a patch for this.
- [x] Figure out a way to allow custom rama_prepro maps to be specified for the patched amino acids.
- [x] Figure out a way to allow custom rotamer libraries to be specified for the patched amino acids.
- [x] Debug the rotamer libraries -- there's something wrong there.
- [x] Add support for lazily loading custom scoring tables.
- [x] For _rama_prepro_.
- ~~Think about thread-safety.~~ <i>Put off to a future pull request. The lazily loaded objects are in ScoringManager, which contains a bunch of stuff that needs to be made thread-safe. It's all in one place, in any case.</i>
- [x] Expand mainchain scoring tables to permit N-dimensional data (for future support for beta-amino acids).
- [x] Add support for specifying the dimension, resolution, and grid offset in the database file (so that we're no longer hardcoding 5-degree bins and whether we have grid points in bin centres or on edges).
- [x] Add support for comments in the database file.
- [x] Comment the database files explaining the columns.<br/>
~~\- Possibly add support for variable-resolution mainchain torsion tables (_e.g._ with finer sampling in alpha-helix and beta-strans wells). This has been talked about in the past, and now might be a time to do it.~~ _Beyond the scope of this pull request, though it should be easier to do now._
- [x] For all of the above, add a MainchainTorsionScoringTable class.
- [x] Support symmetrization of gly/ACHRIAL tables.
- [x] Support polycubic interpolation.<br/>
- ~~Support linear interpolation.~~ <i>Put off to a future pull request, since </i>rama_prepro<i> uses exclusively polycubic interpolation.</i>
- [x] Unit tests for _rama_prepro_ scoring with N-methylation --> RamaPrePro tests AND cyclic geometry tests.
- [x] Unit tests for _fa_dun_ scoring and rotamer generation with N-methylation --> Covered in cyclic geometry tests and protocols::cyclic_peptide::N_methylation tests.
- [x] Beauty.
- [x] Documentation.
- [x] RAMA_PRE_PRO lines in params files.
- ~~Rotamer lines in patch files.~~ --> We have no documentation for patch files. Grr.
- ~~RamaPrePro lines in patch files.~~
- [x] Figure out how to add sampling tables / CDFs for arbitrary mainchain potentials.
- [x] Add function to produce random mainchain torsion vectors biased by probability distribution.
- [x] Unit test.
- [x] Debug why we don't seem to be using the N-methyl rama map currently.
- [x] Add option to use rama_prepro instead of rama tables in GenKIC when sampling.
- [x] Document this.
- [x] Unit test.
- [x] Integration test.
- [x] Add rama_prepro GenKIC filter.
- [x] Document this.
- [x] Add to unit test.
- [x] Add to integration test.
- [x] Add option to GenKIC to correct polymer bond-dependent atoms.
- [x] Document this.
- [x] Turn this on in simple_cycpep_predict.
- [x] Add support for N-methylation to _simple_cycpep_predict_.
- [x] Add option (true by default) to use rama_prepro for sampling instead of rama.
- [x] Document the new option (rama_prepro instead of rama).
- [x] Interface for specifying N-methylated positions.
- [x] Documentation.
- [x] Integration test.
- [x] Use rama_prepro cutoff energy instead of rama cutoff energy.
- [x] Move tryptophan rotamers into database for unit tests.
- [x] Move the rest of the rotamers into the rotamer library.
- [x] Unit test with N-methyl-tryptophan rotamers. --> Covered in protocols::cyclic_peptide::N_methylation tests.
- [x] Move generic N-methyl-L-alpha-amino acid rama table into database.
- [x] Cyclic N-methyl amino acid scoring unit test.
- [x] Regular NMe.
- [x] For beta_nov15 NMe.
- [x] For beta_nov16 NMe.
- [x] Non-NMe beta_nov16 unit test while I'm at it.
- [x] Two-chain NMe.
- [x] Two-chain beta_nov15 NMe.
- [x] Two-chain beta_nov16 NMe.
- [x] Non-NMe beta_nov16 two-chain unit test while I'm at it.
- [x] Mirror image N-methyl amino acid scoring unit test.
- [x] Note to self: @twcraven will regenerate the N-methyl map (and add maps for pre-proline, and for sarcosine and sarcosine pre-pro). I need to add and test those.
- [x] Need sarcosine.
- [x] Need sarcosine_prepro.
- [x] Sarcosine unit tests.
- [x] Fix problem with rebuilding polymer bond-dependent atoms (methyl group).
- [x] Document new option ("update_polymer_bond_dependent_atoms") in ModifyVariantTypeMover.
- [x] Unit test for recursive identification of connection-dependent atoms.
- [x] Fix issue with rebuilding polymer bond dependent methyl group when it's at a terminus.
- [x] Figure out why first output in simple_cycpep_predict_nmethyl integration test has crooked N-methyl group.
- [x] OK, figured it out: it's the packer. When rotamers are constructed, backbone atom coordinates are preserved, by the N-methyl group is not considered to be a set of backbone atoms. Add recursive check for polymer bond dependencies in Residue::place to correct this.
- [x] Add option to MutateResidue mover to update polymer bond-dependent atoms.
- [x] Document this.
Also:
- [x] Updating HydroxylTorsionPotential for D-amino acids.
- [x] Add params files for noncanonical amino acids found in cyclosporin A.
- [x] Add AIBparams #file.
For a future pull request:
- Update _rama_ as _rama_prepro_ was updated.
- Update _p_aa_pp_ as _rama_prepro_ was updated.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #2167 from RosettaCommons/roccomoretti/boron_fix
Fix Boron typing
The atom type used for boron in the sdf/CCD C++ code was mis-specified (it was correct in molfile_to_params.py). Change to the correct type and add a unit tests which covers boron.
Thanks to Greg Sliwoski for pointing out the issue.
notify author
notify list [rosetta-logs@googlegroups.com]