- mmCIF output now has better support for multiple chains, ligands, DNA/RNA
and water, adds chemical component information and user-provided angle
and dihedral restraints, and includes both Modeller ("author-provided")
residue and chain IDs as well as the mmCIF identifiers ('seq_id',
'asym_id').
- Models can now be read from BinaryCIF format files. Templates can also
be provided in BinaryCIF format.
- The default file format for Model.read() has changed from
PDB_OR_MMCIF to PDB_ANY. A file with a '.bcif' extension will now
be read as a BinaryCIF file, one with a '.cif' extension will be read as
mmCIF, while all other files will be read as PDB.
- The internal residue name length has been increased from 4 to 10
characters, to better handle the new 5-character PDB residue names.
- Bundled version of HDF5 updated to 1.14.6.
- Bugfix: atom files are now searched for with the 'pdb' prefix in each
directory specified by IOData.atom_files_directory before moving on to
the next directory; previously this behavior was reversed.
- Bugfix: avoid numerical instability in Model.build() when using
internal coordinates that involve colocated atoms.
- Bugfix: correctly relabel sidechain atoms in chiral amino acids
(VAL/THR/LEU/ILE) to get the correct chirality if necessary.
- Bugfix: author-provided chain IDs and residue numbers are now used when
reading sequence from mmCIF or BinaryCIF, for consistency with reading
coordinates.
