Changes since release 9v7

• Residues treated as BLK no longer have to be HETATM residues; they will be marked as ATOM if so defined in restyp.lib or in the template PDB (for example, DNA or RNA residues). automodel.nonstd_restraints() now builds restraints on both HETATM and BLK residues.

• Python methods that previously expected lists or tuples can now work with other iterables too (such as generators). For example, in Python 2.4 you can now say 's = selection(x for x in mdl.atoms where x.biso > 0.5)' to select all atoms with B${}_{iso}$ over 0.5.

• The file argument to model.write_data() is now optional, since we may only want the calculated property in B${}_{iso}$ , and some properties (e.g. CRV) cannot be written to files anyway.

• Bonds between BLK residues are now detected by a simple distance criterion and are restrained more strongly than other distances by automodel.nonstd_restraints().

• A new method Chain.join() allows chain breaks between chains to be removed.

• Files compressed with 'xz' can now be read, provided that the 'xz' binary is available on your system.

• Bundled version of HDF5 updated to 1.8.3.

• The allow_alternates option to alignment.append() will now allow G in the alignment sequence to match any non-standard residue type in the PDB file which maps to G. This is useful if the alignment sequence is 'cleaned' as by sequence_db.read().

• When using GB/SA with BLK residues, each BLK atom will now be treated as an uncharged particle with a small radius, rather than causing an error.

• In sequence_db.filter(), if two sequences fail to align, don't terminate the entire run; rather, place them in separate clusters.

• density.read() now supports CCP4 files, and reads the map and voxel sizes from the density map file header if not specified by the user.

• Bugfix: invalid PAP files no longer crash Modeller with an assertion failure.

• Bugfix: projected coordinates from environ.principal_components() are now written correctly to the output file.

• Bugfix: models can now be built containing HSE or HSP residues without the program terminating with a cryptic MDT error; homology-derived restraints for these residues will look very similar to those for HIS.

• Bugfix: assigning to Residue.num now gives a correctly-formatted (right-aligned) PDB residue number by default.

• Bugfix: assigning to Residue.code, Residue.name or Residue.pdb_name, where the residue is in an alignment, now works correctly (previously, the new one-letter code was not used when the alignment was written to disk).

• Bugfix: mainchain curvature values for terminal residues were incorrect in some cases.

• Bugfix: mainchain curvature calculation no longer spans chain breaks.

• Bugfix: in profile.scan(), if the target profile has no name, do not trigger a GLib CRITICAL error.

• Bugfix: automodel with write_intermediates=True now works correctly.