Changes since release 9v6

• A new method alignment.get_suboptimals() allows the suboptimal alignment file generated by alignment.salign() to be easily parsed.

• A new method iterative_structural_align() provides a straightforward way to obtain the best SALIGN structure-structure alignment.

• Problems with statistics (e.g. profile.build() on a small database) now raise a StatisticsError rather than the more generic ModellerError.

• Sequence mismatches between PDB files and alignments now raise a SequenceMismatchError rather than the more generic ModellerError.

• alignment.append() and alignment.read_one() now take a new argument, allow_alternates, which allow for alternate matches of one-letter codes when reading PIR sequence ranges from PDB files (B to N, Z to Q).

• An error message is now reported when using an alignment containing an empty sequence, and reading of the full sequence from a PDB file fails.

• Tab characters in alignment files are now treated the same as any other whitespace, and do not result in cryptic failures.

• Parallel jobs have a new method, job.yield_tasks_unordered(), which is like job.run_all_tasks() but which returns results as they are generated rather than when all the tasks have finished.

• Parallel jobs now default to trying to duplicate the current Python environment (using modslave.py and modpy.sh if necessary) when setting modeller_path, rather than running the mod9v7 script, since the latter may conflict with the Python version of the master.

• Bugfix: TER records are now emitted even for single chain proteins, and contain a unique serial number, the correctly formatted residue name, and the chain ID, as per PDB specifications.

• Bugfix: correct slave Python search path when a parallel job master script is run from a directory other than the current working directory.

• Bugfix: profile.build() no longer crashes when searching empty or very small (less than 10 sequences) sequence databases.

• Bugfix: density.grid_search() no longer exhausts the stack when using large density maps.

• Bugfix: PROFILE or BLOCK alignment with alignment.salign() should now work correctly even when align_block=1.

• Bugfix: alignment.malign() and alignment.salign() now work correctly with alignments containing zero-length sequences.

• Bugfix: when reading multiple copies of the same align code from an alignment file, all sequence metadata is now duplicated correctly.

• Bugfix: scores from the multiple binormal restraint were slightly incorrect in some cases.

• Bugfix: initial loop models containing hydrogens are now read back in correctly (fixes an 'Index x out of range' error).

• Bugfix: residues marked in PDB file REMARKs as BLK are preserved as BLK (rigid bodies) when the PDB is read back in, even when their name appears in the topology library.

• Bugfix: PDB RESOLUTION lines should now be parsed correctly in both old-style (v3.1 or older) PDB files and new-style (v3.2) PDBs.