Changes since release 9v3

• The speed of GB/SA calculations has been dramatically increased; the cutoff distance used in the GB/SA calculation can also now be adjusted to balance speed against accuracy (the default cutoff has been decreased from 15.0Å to 8.0Å).

• alignment.append() no longer takes rewind_file and close_file arguments; they did not work correctly anyway. New method alignment.read_one() added to read sequences one by one from a PIR or FASTA file.

• alignment.write(), model.write(), selection.write(), Chain.write() and Structure.write() now take either a file name or a file handle (modfile.File()).

• Split alignment.check() into alignment.check_structure_structure() and alignment.check_sequence_structure(). Both functions now return the number of pairs which exceeded the cutoffs, which are also adjustable.

• alignment.compare_with() now returns the percentage of equivalent residue-residue pairs in the two alignments.

• io_data.atom_files_directory is now a Python list of directories, rather than a colon-separated string. (This is so that everything works on Windows, where ':' is a valid character in directory names.) Old scripts that use colon-separated strings should still work, but will trigger a warning.

• When reading or writing compressed files, the uncompressed versions of these files are now written into writeable local storage (${MODELLER_TMPDIR} if set, otherwise the system temporary directory if possible, otherwise the current working directory). Previously, these files were written into the same directory as the compressed versions, which is problematic when that directory is read-only or is on network storage.

• Files compressed with p7zip can now be read, provided that the 7za binary is available on your system. Such files should contain only a single member (e.g., a file foo.pdb could be compressed into the archive foo.pdb.7z).

• Update PDB files used in the examples (remediated PDBs are now used).

• Added selection.only_water_residues() method, which selects only atoms in water residues.

• User-defined energy terms now take a cutoff parameter to use the non-bonded list.

• Bugfix: use the basename, not the full path name, to generate output file names in density.grid_search().

• Bugfix: model.transfer_xyz() with cluster_cut > 0.0 incorrectly calculated the number of atoms in each residue, which in some circumstances could cause a crash.

• Bugfix: miscellaneous alignment.salign() fixes.

• Bugfix: alignment.append_model() now copies the model's resolution, R-factor, name and source to the alignment.

• Bugfix: if two or more different types for a single residue are listed in a PDB with alternate location codes (e.g. 80:A in 1en2), only the first is read by MODELLER.