Use mmCIF res->seq correspondence info

[Greg Couch]

Doesn't matter if it is mmCIF or PDB format. Traceback:

(gdb) where
#0  0x00007fffdb965b5c in std::vector<char, std::allocator<char> >::size (this=0x0)
    at /home/socr/a/chimera/include/c++/6.3.0/bits/stl_vector.h:656
#1  0x00007fffd6d0df46 in atomstruct::_constrained (aseq=..., cm_ap=..., offsets=..., max_errors=0) at seq_assoc.cpp:172
#2  0x00007fffd6d0ea69 in atomstruct::_constrained (aseq=..., cm_ap=..., offsets=..., max_errors=125) at seq_assoc.cpp:285
#3  0x00007fffd6d0e909 in atomstruct::_constrained (aseq=..., cm_ap=..., offsets=..., max_errors=268) at seq_assoc.cpp:273
#4  0x00007fffd6d0e47d in atomstruct::_constrained (aseq=..., cm_ap=..., offsets=..., max_errors=268) at seq_assoc.cpp:241
#5  0x00007fffd6d0eee2 in atomstruct::constrained_match (aseq=..., mseq=..., ap=..., max_errors=268) at seq_assoc.cpp:321
#6  0x00007fffd6d0f951 in atomstruct::try_assoc (align_seq=..., mseq=..., ap=..., max_errors=268) at seq_assoc.cpp:439
#7  0x00007fffd6ca8c86 in atomstruct::AtomicStructure::make_chains (this=0x10f2290) at AtomicStructure.cpp:380
#8  0x00007fffdb961478 in atomstruct::Structure::chains (this=0x10f2290)
    at /home/socr/a/gregc/src/chimerax/build/include/atomstruct/Structure.h:206
#9  0x00007fffdb961502 in atomstruct::Structure::num_chains (this=0x10f2290)
    at /home/socr/a/gregc/src/chimerax/build/include/atomstruct/Structure.h:244
#10 0x00007fffdb96e9bf in error_wrap_array_get<atomstruct::Structure, unsigned long, unsigned long> (instances=0x7fffd43104d8, n=1,
    pm=(unsigned long (atomstruct::Structure::*)(const atomstruct::Structure * const)) 0x7fffdb9614ea <atomstruct::Structure::num_chains() const>, args=0x7fffde06f560) at molc.cpp:186
#11 0x00007fffdb959529 in structure_num_chains (mols=0x7fffd43104d8, n=1, nchains=0x7fffde06f560) at molc.cpp:4002
#12 0x00007fff4b20112a in ffi_call_unix64 ()
    at /home/socr/a/gregc/src/chimerax/build/tmp/Python-3.6.3/Modules/_ctypes/libffi/src/x86/unix64.S:76

[Greg Couch]

Also caused by 2ftc.

[Greg Couch]

And 3jbr

[Greg Couch]

And 4v47

[Greg Couch]

and 4v48

[Greg Couch]

and 4v5z

[Eric Pettersen]

Fixed 4adc/2ftc crashes by making nucleic chain-trace connectivity cutoff more permissive.

[Eric Pettersen]

For some recursion paths, the association parameters were not being initialized (fixes 3jbr).

[Greg Couch]

Would like an API that would let the mmCIF reader set the residue location in the sequence to bypass any recursion.

[Eric Pettersen]

Yes, that's the plan.

[Eric Pettersen]

Would you want to specify the correspondence info as part of the set_input_seq_info() call, or separately in a later call?

[Eric Pettersen]

Awaiting feedback from Greg

[Greg Couch]

The seq_input_seq_info() call would work. I have all of the information at that time.

[Eric Pettersen]

Okay, the API is implemented.

Structure.set_input_seq_info now takes two additional (optional) arguments: vector<Residue*> [corresponding residues], and PolymerType. If either is given then both must be provided. The corresponding-residue vector has to be the same length as the residue-name vector, and therefore is allowed to contain null pointers.

Ball's in your court for now.

--Eric

[Greg Couch]

Implemented for mmCIF files. Fixes 2adw. Needs to be done for mmtf files too.

[Eric Pettersen]

Speaking of MMTF -- it needs to call use_best_altlocs once it has created the structure

[Greg Couch]

Done.