Merge pull request #307 from RosettaCommons/rfalford12/default_hbonding_and_copying
Rfalford12/default hbonding and copying
======================================
Small updates to the membrane framework:
= Membrane hydrogen bonding will turn on automatically - no additional flags needed
= Fixed issue where MembraneInfo was incorrectly copied in Conformation
= Cleaned up unused variables and interfaces in AddMembraneMover
= Updating relax and symmetric docking integration tests to reflect current protocol state (same as 2015 demos)
Expected Integration test changes: mpframework_symdock and mpframework_relax
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #303 from RosettaCommons/vmullig/bundlegridsampler_resetmode
Adding a reset option to the BundleGridSampler.
This just adds one option to the BundleGridSampler. No unit or integration tests expected, except for a trivial cosmetic change to the BundleGridSampler output in its integration test.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #298 from RosettaCommons/revert-293-revert-280-jkleman/mp_updates
Revert "Revert "Jkleman/mp updates""
Merging in the stuff from an earlier reverted commit:
= corrected MPDocking weights from incorrect weights and added smooth terms and weights: mpframework_docking_<fa/cen>_2014.wts; these should be used from now on
= changed instances from using OP as args to &
= AddMembraneMover:
* switched SSize for membrane residue to regular Size
* spanfile is not a required input any more: if flag -membrane_new:setup:spans_from_structure <bool> is passed, it calculates the topology from the structure => this requires the structure to be transformed into membrane coordinates
* reading in MEMs got better: AddMembraneMover will automatically detect MEMs without a flag; flag can be used if multiple MEMs are present
* one of the constructors can now also use the anchor residue as input, default is 1 throughout
= SpanningTopology:
* adjusted method fill_from_structure so that it can be used from outside the class
= compute_structure_based_membrane_position:
* renamed to compute_structure_based_embedding
* overloaded functions to also take SpanningTopology as input: this is more useful for Symmetry or computing embedding for parts of the pose where the topology isn't necessarily for the whole pose
= protocols/membrane/geometry/util
* added split_topology_by_jump method and unit test for it
= span_from_PDB:
* changed default thickness from 12.5A to 15A
= core/pose/util.hh
* added function nres_protein that counts protein-only residues in the pose
= MEM / EMB params files:
* made all of them consistent to match the fa MEM params file, numbers didn't change except thickness is now consistently 15A for both low-res and high-res
* order of atoms in EMB changed, but this shouldn't affect any calculations
= MPDockingMover:
* reorganized code to be able to add stuff easier
* constructor now takes a jump number for docking
* added option to read in scorefunctions for both low-res and high-res steps
= FlipMover:
* added and unit tested: flips a pose element in the membrane, given a jump number
Unit tests: pass
Integration test changes: Tracers for Membrane framework are fine.
Integration.release_debug changes: Known numerical instabilities
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #300 from RosettaCommons/phbradley/lk_ball_update
Updates to the lk-ball energy term, to improve compatibility with minpack and linmem ig. Integration test changes due to extra core.chemical Tracer output (new atom properties associated with the lk_ball_wtd term). Two broken cenrot unit tests were already broken in trunk, maybe by the recent lbfgs checkin?
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #297 from RosettaCommons/vmullig/layerdesign_ncaas
Tweak LayerDesign so that it can work with noncanonical residue types
This alters the LayerDesign TaskOperation so that it can work with noncanonical residue types. In addition, it removes highly counter-intuitive behaviour in LayerDesign that breaks TaskOperation commutativity (while adding an option to turn that behaviour back on if that's what the user really wants).
I'm also turning off "DCYD.params" and "DCYS.params" in residue_types.txt. These D-amino acid residues should not be loaded by default, and can be specified with the -extra_res_fa flag.
Tasks:
- Stop having LayerDesign skip residues that were set with PIKAA or NATRO in a resfile as the default behaviour. This is counter-intuitive behaviour for this TaskOperation because the skip ONLY occurs if the resfile was applied before LayerDesign -- i.e. resfile,LayerDesign is different than LayerDesign,resfile, which violates the rule that TaskOperations are commutative.
- Add an option to restore the old, non-commutative behaviour if an advanced user so desires (while keeping it off by default for users who need to be able to assume that TaskOperations follow consistent rules).
- Stop automatically restricting non-protein residues to repacking. If the user wants to do this, he or she can do so with another TaskOperation. TaskOperations should be atomistic; they shouldn't do several unrelated things that limit their usefulness.
- Add code to LayerDesign to parse NCAA options.
- Add code to LayerDesign to actually add NCAAs to the PackerTask.
- Alter the use_residue_neighbors code so that it uses the (first sidechain atom)-(first sidechain atom parent) vector rather than the CB-CA vector.
- Add integration tests for NCAA design with LayerDesign.
- Update the documentation wiki.
Expected unit test failures: NONE.
Expected integration test changes:
- LayerDesign (trivial, cosmetic change to output).
- LayerDesign_ncaa (newly-added test).
- fibre_diffraction_fad (cosmetic changes to output since the DCYD and DCYS params files are no longer being loaded).
- mm_params (cosmetic changes to output since the DCYD and DCYS params files are no longer being loaded).
- fold_and_dock (cosmetic changes to output since the DCYD and DCYS params files are no longer being loaded).
- number_of_residuetypes (since the number of residue types is changing, since DCYD and DCYS are no longer being loaded).
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #278 from RosettaCommons/rhiju/trace_assert
Rhiju/trace assert
only integration test changes are the ones that are currently broken.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #284 from RosettaCommons/lqtza/pyrosetta_pymol_energy_labels_fix
PyRosetta: Fixed bug for sending energies from pyrosetta to pymol as ca-labels
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #275 from RosettaCommons/JWLabonte/utilities
A few years back, I added what I consider a couple helpful convenience methods to vectorL. I had never written unit tests for them; now I have.
This merge also introduces a third convenience method for appending one vector into another, which would be really helpful for some sugar code I'm working on.
There is a simple carbohydrate code fix as a rider to this merge, so carbohydrate integration test changes are expected. All other tests pass.
notify author
notify list [rosetta-logs@googlegroups.com]