MantisBT - Issues

0000366: Centroid mode docking doesn't work with Nucleic Acids?

Thu, 28 Apr 2016 18:36:48 -0500

Forum links:
https://www.rosettacommons.org/content/no-output-pdb-files-after-docking-protein-and-dnarna [^]
https://www.rosettacommons.org/node/9603 [^]

Apparently, there is/was an issue with using the docking protocol with nucleic acids, as the centroid mode docking code and centroid mode scoring has a bunch of "if (rsd.is_protein())" checks.

I (Rocco) have not personally checked this, I'm just forwarding a bug from the forums so it doesn't get lost. Note that the second forum post links to a (non-RosettaCommons lab) paper which contains a patch for protein/nucleic acid docking

0000365: Centroid mode doesn't work reliably for hydroxylproline

Thu, 28 Apr 2016 16:59:28 -0500

See https://www.rosettacommons.org/content/creating-centroid-patches-proline-prohydroxylatedcase1-and-prohydroxylatedcase2 [^]

I'm opening the bug to record the suggested fix:

> Line 0000002 of the pro_hydroxylated_case1.txt and pro_hydroxylated_case2.txt
> files located in /centroid/patches reads "TYPES HYDROXYLATION".
> Presumably, this is incorrect. It should be "TYPES HYDROXYLATION1" for
> pro_hydroxylated_case1.txt and "TYPES HYDROXYLATION2" for
> pro_hydroxylated_case2.txt.

0000364: C-terminal_conjugation patch crashes several unit tests

Tue, 12 Apr 2016 14:15:01 -0500

When the C-terminal_conjugation patch is available to Rosetta, 13 unit tests fail, some of them crashing and all having to do with packing &/or rotamer-generation, from the following suites:

AACompositionEnergyTests
MirrorSymmetricConformationTests
AntibodyTaskOps
ConsensusLoopDesignOperationTests
DDGScan
RestrictOperationsTests

See also: https://github.com/RosettaCommons/main/pull/1259 [^]

0000363: rogue PDB output in core/scoring/packing/PoseBalls.cc

Tue, 29 Mar 2016 12:19:30 -0500

"rogue PDB output located in core/scoring/packing/PoseBalls.cc

it puts, uh, the balls in the PDB output. so I guess they should be appended sorta the way MEM or orbitals are, like ATOM records."

(From Andy W)

0000361: accumulation of REMARK output in packRotamersMover when nstruct>1

Mon, 28 Sep 2015 12:26:46 -0500

The PackRotamersMover tracks and writes to the output pdb the residues which are packed or designed using REMARK lines.
When nstruct > 1 the REMARK lines which track packable and designable residues and write out to the output pdb accumulate. So, if you use the PackRotamersMover 3 times then the number of REMARK lines is 3 times the number of decoys generated by the currently running process.

The issue tracks to PackRotamersMover::note_packertask_settings. Bug identified in 2015.05.57576 but code has not changed since 2011.

0000190: -jd2:delete_old_poses and JD2 memory leak

Tue, 18 Aug 2015 11:13:05 -0500

JD2 has a longstanding memory leak that becomes quite serious in a large -l environment. "Old poses" - inputs that Rosetta is through with and will not reuse - are maintained in memory.

The flag -jd2:delete_old_poses prevents this memory leak but affects Oliver's code for reloading results into the JobDistributor as new jobs, and may affect RosettaScripts due to modification of Poses before distribution starts.

0000007: passing PDBs to -l causes unexpected behavior

Tue, 18 Aug 2015 03:07:45 -0500

If you pass a PDB to -l instead of to -s, Rosetta accumulates gobs of memory (10GB+) before your sysadmin gets mad and kills the job. It would be useful to track down why it is doing this...

0000359: out:path:all does not work for features/database output

Tue, 18 Aug 2015 00:53:34 -0500

out:path:all works for score files and PDBs, but not features databases. The new databases always go to the current working directory.

0000140: FoldTree::reorder fails to reorder perfectly good fold trees

Tue, 14 Jul 2015 09:50:25 -0500

FoldTree::reorder fails to reorder perfectly good fold trees sometimes. I can't tell if there's a legitimate reason they can't be reordered, or if the function is just failing. There are a few reorder failures in the integration tests (all in my UBQ_E2 code), but they seem not to be a problem (it is failing to reorder(1) when root was already 1).

0000360: database sqlite output options don't do anything if using features reporter in rosetta scripts

Thu, 18 Jun 2015 00:26:17 -0500

-database_partition does not append the database name nor create a partitioned schema. However, the tag in the xml

>

does produce a paritioned schema

0000358: new etables keep being created in optE

Mon, 09 Mar 2015 19:35:22 -0500

OptE started chewing up a lot of memory when the analytic etable was not being used; a quick bit of cout debugging showed that a new etable was getting created for each design job. Something fishy is going on in the way etables are retrieved from the scoring manager. The likely culprit is the EtableOptions::operator ==.

0000353: "did you forget to pass -overwrite?" printed even if no jobs are considered.

Fri, 28 Nov 2014 15:18:04 -0600

In JD2, the "did you forget to pass -overwrite?" help message is printed even in cases where it wouldn't actually do anything - for example, if there weren't any jobs even considered.

0000357: Lei Shi has duplicated almost the entirety of BackboneStubConstraint in creating BackboneStubLinearConstraint

Fri, 21 Nov 2014 08:31:44 -0600

Code duplication.

0000356: Minirosetta threading protocol segfaults on alignments with no gap.

Tue, 04 Nov 2014 11:12:04 -0600

If you use the threading protocol on the minirosetta application with an alignment which doesn't have any gaps, Rosetta will segfault. The error occurs in protocols::loops::fold_tree_from_loops(), where root() is taken from an empty FoldTree. The reason for the empty FoldTree is that the loops passed in is empty, because in LoopRelaxMover::apply(), pick_loops_chainbreak() returns no loops.

0000355: backrub application and JD2 integration

Mon, 03 Nov 2014 18:01:13 -0600

The backrub application was converted to JD2, but doesn't take advantage of the JD2 output features, making it difficult to target alternative output formats (e.g. silent files). Additionally, the output structure that is chosen is the lowest energy structure, when the common usage of backrub apparently makes use of the last structure instead.

It might be nice to use the auxiliary output handling of JD2 to deal with the output of the lowest energy structure in addition to the output of the last structure by the common route.

0000354: Missing residues from ROSIE antibody modeling

Tue, 21 Oct 2014 12:58:01 -0500

A user used the antibody modeling apps through ROSIE and had two residues missing in the final antibody structure.

https://www.rosettacommons.org/node/3769 [^]

From conversations with people more knowledgeable than I in regards to the antibody suite, it most likely has to do with the framework that was chosen from the input sequence.

In cases like these, the antibody modeling script could call the partial_thread application and then the hybridize comparative modeling application to create the full structure of the antibody without the CDRs. This may be difficult as hybridize may try to connect the CDR stems. Perhaps a call to Remodel would be sufficient to add the missing residues.

0000350: Fragment Picker SegFaults when not given a sequence

Mon, 22 Sep 2014 11:41:10 -0500

Report from Forum post (https://www.rosettacommons.org/node/3797 [^])

A user tried to run fragment picking by using the "-s" option with a PDB, rather than using any of the sequence input options. This resulted in a segfault in protocols::frag_picker::scores::MakeSecondarySimilarity::make(), presumably when FragmentPicker::get_query_seq() is called, returns a null OP, and is dereferenced.

0000349: Linux versioning issue with PyRosetta BuildBindings.py

Wed, 13 Aug 2014 13:02:10 -0500

When getting the version of Linux BuildBindings.py apparently assumes a single digit point release:

lib_path = 'build/src/'+mode+'/linux/' + platform.release()[:3] + '/' + PlatformBits +'/x86/gcc/'"

This causes a mismatch issue with scons's version parsing.

0000347: Seqpos in ProteinResidueConformation

Wed, 30 Jul 2014 01:13:31 -0500

Most of the features reporter use resNum to define the Rosetta numbering of amino acids in a pdb; however ProteinResidueConformation uses seqpos. Ideally this nomenclature should be consistent.

0000345: The rama score check in CCD closure behaves exactly opposite the way it should.

Tue, 29 Jul 2014 16:37:20 -0500

If the rama check flag is set to true during CCD closures, the moves are always accepted if the rama check score is WORSE!

0000039: Running docking on a single-chain PDB causes a segfault

Sat, 19 Jul 2014 13:50:54 -0500

Running docking on a single-chain PDB causes a segfault

0000332: Problems compiling with uint64_t

Sat, 19 Jul 2014 13:23:58 -0500

On some platforms it appears that the definition of uint64_t is missing/faulty.
This expresses itself as a compilation error in src/protocols/wum2/EndPoint.cc:

src/protocols/wum2/EndPoint.cc:21:42: error: a function call cannot appear in a constant-expression

If you supplement the definition of uint64_t with the one defined by boost/cstdint.hpp this apparently fixes the error.

0000278: AtomPairConstraint does not check existence of atoms (?)

Sat, 19 Jul 2014 10:55:00 -0500

This is a bug I've run into in old versions of Rosetta (3.4 and earlier). A casual reading of trunk suggests this has been fixed but we need to check it.

AtomPairConstraint does not appear to check if atoms actually exist before attempting to apply constraints. If you pass fullatom constraints into centroid mode, where the sidechain atoms do not exist, the constraint functions look at garbage atom positions and return garbage constraint scores. Occasional crashes occur from segfaults, but usually it looks at allocated but inappropriate memory.

Here's a 3.4 example. https://www.rosettacommons.org/node/3278 [^]

I have not yet attempted to reproduce this in trunk.

0000346: Using IntefaceAnalyzer with a ligand and -fixedchains results in segfault

Mon, 07 Jul 2014 14:12:12 -0500

In a two-protein, one small molecule ligand protein, using the -fixedchains option (either with "-fixedchains A B" or "-fixedchains X") results in a Segfault. This is likely due to the single residue peptide edge made in the foldtree:

apps.public.analysis.InterfaceAnalyzer: Fixed chains are: A, B, these will be moved together.
core.kinematics.FoldTree: FoldTree::reorder( 287 ) failed, new/old edge_list_ size mismatch
core.kinematics.FoldTree: 5 4
core.kinematics.FoldTree: FOLD_TREE EDGE 1 234 -1 EDGE 1 287 1 EDGE 235 286 -1 EDGE 235 234 2 EDGE 287 287 -1
Segmentation fault

0000343: Options file do not use -in:path directory locations

Thu, 26 Jun 2014 14:46:49 -0500

Options file loading should use set paths for locating files.