Changes since release 9v8

• Modeller is now only supported on Windows XP Service Pack 2 or later. If you want to run Modeller on Windows ME, 2000, or NT, use Modeller 9v8.

• Added support for Python 2.7; Python 2.5 is now supported on all platforms.

• Bundled version of HDF5 updated to 1.8.5-patch1.

• Initial models generated by automodel now always contain valid PDB coordinates, i.e. each coordinate is in the range -999.999 to 9999.999.

• During loop modeling, if atoms are missing from the initial model, loopmodel.loop.write_defined_only can be set True to write out PDBs containing only non-loop atoms present in the initial model.

• Restraints using user-defined one-dimensional mathematical forms can now be converted to splines using Restraints.spline().

• Bonds between protein atoms and BLK residues are now restrained more strongly than regular BLK-Calpha distances.

• alignment.salign() now generates the requested number of suboptimal alignments, not one less.

• When reading new-style PDB files (with element names in columns 77-78) Modeller will now use this information to determine whether an atom is a hydrogen. (With old-style or non-conformant PDB files, it will continue to guess based on the atom name, so will not read mercury ions for example unless env.io.hydrogen=True, since it cannot unambiguously determine whether HG is mercury or "hydrogen bonded to CG".)

• The Windows installer now supports silent installation with the /S flag. Use the /D flag to specify install location and /KEY to specify the Modeller license key, e.g. setup.exe /S /KEY=xxx /D=C:\Program Files\modeller

• Bugfix: complete_pdb() now preserves chain IDs.

• Bugfix: environ.dendrogram() no longer crashes if given a distance matrix of size zero.

• Bugfix: Random seed is no longer ignored by model.make_region().

• Bugfix: MET residues marked as HETATM (e.g. ligands) are now correctly skipped when env.io.hetatm=False (previously they were read in).

• Bugfix: lookup of atoms by name in models containing duplicated residue and/or chain names now works correctly.

• Bugfix: fix PSA crash on models containing no amino acids.

• Bugfix: alignment.get_suboptimals() no longer fails to parse alignments where neither sequence starts at the N terminus, and now correctly handles alignments where the C termini are misaligned.

• Bugfix: sequences in PAP files can now start in any column (as long as all the sequences start in the same column), not just column 11.

• Bugfix: allow atom names in the group_restraints atom class file to contain single-quote characters (e.g. nucleic acid atoms such as O3').

• Bugfix: fix range of spline math forms when resplined by Restraints.spline().