merge of master with JK-lab after fixing the issues with reading opencl files in MAC os
OpenCL driver (AMD) was crashing with seg faults while reading the input opencl files from Rosetta. This happens only in Mac OS. Now, RegisterProgram function in GPU in rewritten to fix the problem.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #226 from RosettaCommons/rhiju/ester_oxygens
creating option to allow testing of turning off ester hydrogen bonds.
All integration test changes are trivial, from changed atom type numbers, or in tests that are currently problematic (kic_with_fragments)
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #220 from RosettaCommons/JWLabonte/sugars/output
Carbohydrates: GWS file output
This merge introduces GlycoWorkbench structure (GWS) file output to Rosetta for carbohydrate chains.
This is the standard string format used by the glycobiology community, and the files can be opened in the GlycoWorkbench software to give a 2D "polygon" topology diagram of glycans. The files can also be linked to various glyco databases.
Future work will add GWS-reading functionality.
This merge also includes a miscellany of other sugar-related stuff from other branches that I merged into this branch because I needed them, such as:
* outputting `LINK` records to `.pdb`s
* more sugars in database
* more sugar patches in database
* more lipids in database
* sugar docking scripts/pilot apps
* code clean-up of some of my older code
* output improvements to rigid-body docking protocols
* clean-up of `pilot_apps.src.settings.all`
All unit tests pass.
`carbohydrate` integration test changes expected.
Ligand-docking and related integration test changes also expected, due to a minor improvement to RigidBodyDocking output.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #212 from RosettaCommons/jkleman/compute_membrane_position
Jkleman/compute membrane position
**Updates for MembraneFramework: compute_structure_based_membrane_position**
* protocols/membrane/geometry/util:compute_structure_based_membrane_position: removed code duplication for this function; now this function uses the constructors in Embedding that takes a pose and a topology object
* Embedding: Embedding from span now checks for the dihedral angle between the first span and each consecutive one to calculate embedding normal instead of positive/negative z-coordinate or odd/even spans; this is much less error-prone and works for multiple chains. The dihedral angle cutoff is 100 degrees. Works for the cases I looked at but is not extensively benchmarked.
* membrane normal: normalized membrane normal vector to length 15 instead of 1 for better visualization and debugging purposes; same holds true for embedding normal vectors
* VisualizeEmbeddingMover: added VisualizeEmbeddingMover; when using this mover embedding virtuals are added to the pose after the membrane residue. Can be invoked using the flag -membrane_new::visualize::embedding <boolean>
* EmbeddingDef / Embedding: added constructor to create Embedding object from a single EmbeddingDef object
* EmbeddingDef: added function in EmbeddingDef:from_span_positive_z to distinguish calculation of the normal vector into positive z-coordinate instead of any direction
* protocols/membrane/geometry/util: added function average_antiparallel_embeddings which inverts the normals that are showing in the opposite directions; the average_embeddings function just averages which can result in zero vectors if some of them are antiparallel
* TransformIntoMembraneMover: for fixing the unit tests I needed to set the membrane residue at the root in the apply function
* EmbeddingDef / Embedding: added copy constructor and assignment operators
* Embedding: added error checking for embedding(span_number)
* PymolPyRosettaServer: adjusted membrane thickness to work with normals of length 15, removed RG dependence for membrane
**Tests**
Membrane unit tests: pass
Membrane integration tests: expected changes in mpframework_relax and mpframework_loadtime due to different calculation of membrane normal
Other integration tests: all other changes are typically broken tests or numerical changes
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #217 from RosettaCommons/rhiju/fix_binary_protein_silentstruct_full_model_parameters
get rid of full_model_info setup by default in pose_from_pdb
integration test changes either remove FULL_MODEL_PARAMETERS tag from silent files (as desired) or are in tests known to be numerically sensitive (kic_with_fragments) or previous fails.
notify author
notify list [rosetta-logs@googlegroups.com]