Merge pull request #1356 from RosettaCommons/weitzner/ab_auto_kink
Antibody: auto kink! adding the ability to generate a kink constraint on-the-fly for the H3 modeling phase of antibody_H3
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1362 from RosettaCommons/vmullig/move_fastdesign
Moving FastDesign from devel to protocols.
FastDesign is very widely used, and is pretty feature-complete (I don't think anyone is developing it much), but it has been sitting in devel for some time. I want to write a protocols-level algorithm that invokes it, but can't as long as it's in devel, so I'm going to move this to protocols.
Tom Linsky has also gone through this and cleaned up the FastDesign code a bit, removing some extraneous stuff.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1363 from RosettaCommons/jadolfbr/make_cdr_cluster_ident_public
Promote the identify_cdr_clusters app to public, as suggested by @lqtza. This app can be used for any numbering scheme and cdr definition. It will print the cdr cluster information and echo the info to any output PDB.
Documentation is here:
https://www.rosettacommons.org/docs/wiki/application_documentation/antibody/CDR-Cluster-Identification
An integration test is currently in Rosetta.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1360 from RosettaCommons/vmullig/countpair_issue
Cache FA_Elec data that are loaded repeatedly with beta_nov15
With the beta_nov15 scorefunction, certain FA_ElecEnergy data are loaded from disk every time the term is instantiated or cloned. This is a HUGE problem in protocols that copy scorefunctions in loops -- especially when they're being run on Blue Gene. This pull request aims to allow these data to be read once per process and cached as global (static) data.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1358 from RosettaCommons/vmullig/design_cycpep_app
Add filtering for oversaturated hydrogen bond acceptors to simple_cycpep_predict app.
This allows the simple_cycpep_predict app to automatically throw away sampled conformations in which more than the allowed number of hydrogen bonds are being formed to a hydrogen bond acceptor. It pains me to split that infinitive.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1349 from RosettaCommons/vmullig/oversatisfied
Add a filter for hydrogen bond acceptors with too many hydrogen bonds
Rosetta's scorefunction poorly accounts for cases in which too many hydrogen bond donors are trying to bond to a single hydrogen bond acceptor (e.g. three amide protons hydrogen bonding to a single carbonyl oxygen). This is largely because of the pairwise-decomposible hydrogen bonding score terms, which can't be aware of any other interactions the residues in question are making with a third body. This filter is intended to flag these cases.
Tasks:
- Create the .cc, .hh, .fwd.hh, and Creator.hh files (using Jared's handy script).
- Write the parse_my_tag:
- Private member var (core::Size), initialization, getter, and setter for allowed number of oversaturated acceptors.
- Private member var (bool), initialization, getter, and setter for whether we're just counting backbone.
- Private member var (core::Real), initialization, getter, and setter for the energy threshold for counting something as an h-bond.
- Private member vars (ResidueSelectorCOPs), initialization, getter, and setter for two optional selectors for the hydrogen bond acceptor residues and hydrogen bond donor residues.
- Write the apply() function.
- Peripheral functions -- report() and whatnot.
- Unit test.
- Debug.
- Integration test.
- Documentation.
- Beauty.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1355 from RosettaCommons/vmullig/update_rosettascripts_talaris2014
Switching some more defaults to talaris2014 from talaris2013.
Certain utility functions in RosettaScripts still default to talaris2013, and some default to score12. Updating to "commandline", so that they always stay in sync with whatever the default Rosetta behaviour is.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1329 from RosettaCommons/vmullig/peppredict_blugene
Modifying the simple_cycpep_predict application to play nicely on Blue Gene.
I need to do this to allow it to be run on large-scale clusters while BOINC is tied up during CASP. This will also be my test-bed for writing a multi-level job distribution and collection hierarchy (and, ulitimately, adding multi-threading to it). It does NOT replace jd3 in any way; on the contrary, it's a convenient place to test features that we might want in jd3.
Tasks:
- Add -cyclic_peptide:MPI_processes_by_level flag to let the user specify how many processes will be devoted to each level of the hierarchical communications scheme.
- Add option to control size of batches being distributed at each level.
- Modify the simple_cycpep_predict app to set up MPI options.
- Add a SimpleCycpepPredictApplication_MPI class to the protocols level that handles the multi-level communications.
- Revise the job buffer size to a list of job buffer sizes, and remove the requirement that the job buffer size be at least the number of slaves per first-level master.
- Add hooks to SimpleCycpepPredictApplication to let it be set up by SimpleCycpepPredictApplication_MPI, and to suppress its disk I/O.
- Finish implementing all of the functions marked //TODO WRITE ME in SimpleCycpepPredictApplication_MPI.cc.
- Have this class actually distribute jobs down the hierarchy.
- Have the slave nodes call the SimpleCycpepPredictApplication class for sampling.
- Option to return only the top n% of jobs. (Job return is two-step: first, send list of scores up the hierarchy to the emperor; have emperor determine and request top n%; second, send only top poses up the hierarchy.)
- Send results summaries UP the hierarchy.
- Sort sent results as they propagate up the hierarchy.
- Unit tests for sort functions.
- Send requests for poses DOWN the hierarchy based on sorted results.
- Have the SimpleCycpepPredictApplication_MPI class collect results UP the hierarchy.
- Write out silent structures from the emperor node.
- Add integration test to MPI test suite.
- Documentation.
- Test on jojo.
- Test on Blue Gene.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1354 from RosettaCommons/smlewis/Extender_pilot_app_bitrot
trival bitrot repair on a senescent pilot app (CAP->COP); This is a noncompiling pilot app that was useful for generating starting structures for FloppyTail. As a noncompiling app it didn't get fixed when the smart pointers were updated and CAPs went away. This app remains disabled in scons (although it does compile now), so there oughtn't be test changes associated with it (and so I'm not loading the test server with it, either).
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1352 from RosettaCommons/JWLabonte/sugars/docking
This merge improves a private pilot app I used for glycan docking. It also includes the following fixes/changes:
more sugar patches and .params
2-deoxy-
another rhamnose
more H-placement fixes
adding output to notify a user if input rings are not chemically believable
fixes rotamer sets for substituted sugars
(I had planned to add ->6-beta-D-Manp, but it looks to be already in the database....)
All tests pass.
Integration test changes expect because of .params file/patch changes and/or new output.
notify author
notify list [rosetta-logs@googlegroups.com]