Changes since release 9v4

• Default gap_penalties_1d for alignment.align2d() have been changed to match the optimized 2D penalties, as determined from a test alignment set.

• automodel now constrains any protein-ligand close contacts (less than 2.3Å) to attempt to preserve protein-ligand chemical bonds.

• New normalized DOPE assessment method, which returns a Z-score, and is thus comparable between different proteins (see model.assess_normalized_dope()).

• Bundled version of HDF5 updated to 1.8.1.

• PDB atom files are now also looked for in PDB-style subdirectories (the first two characters after the digit); e.g. 1abc is searched for in the 'ab' subdirectory, pdb4xyz.ent in the 'xy' subdirectory.

• Libraries updated to correctly read and write remediated RNA and DNA, some other HETATMs (e.g. ions) and to use the new hydrogen names, in PDB v3.

• Modeller now outputs new-style PDBs, with the element in columns 77 and 78, rather than older-style PDB which has a sequence number in this position.

• Enable Large File Support, to correctly read files bigger than 2GB on 32-bit Unix systems.

• Remove unused residue_grouping and distance_atoms arguments from Restraints.make().

• Bugfix: no longer ignore excluded pairs between two BLK residues.

• Bugfix: selection.only_mainchain() now considers OXT atoms to be mainchain.

• Bugfix: alignment.append_model() no longer crashes when trying to add a structure containing only one atom.

• Bugfix: Topology.make() should now preserve complete IC/BILD entries for patch residues.