Changes since release 9v1

• Added loopmodel.user_after_single_loop_model() method.

• Added loopmodel.loop.library_schedule and loopmodel.loop.max_var_iterations members, to allow the loop modeling optimization protocol to be modified.

• 'parallel' module should now work on all platforms, provided Python is available.

• sge_qsub_job() now groups all automatically-spawned qsub jobs into SGE task arrays.

• dope_loopmodel's confusingly-named initial_model argument is now removed. To provide an initial loop conformation, override the loopmodel.build_ini_loop() method.

• By default, automodel now only randomizes starting coordinates for residues with defined topology (previously it also randomized BLK residues, and in some cases the optimizer was then unable to recover their rigidity). To override this behavior, set automodel.rand_method.

• alignment.salign() now takes a rms_cutoff argument with a single RMS value, rather than an 11-element 'rms_cutoffs' list, since only the first value from this list was used, and alignment.compare_structures() now takes a single varatom argument rather than a 2-element 'distance_atoms' list, for the same reason.

• All relevant data (initial RMS plus cutoff values) are now returned from selection.superpose().

• The API has been cleaned up and given a much more logical structure.

• Windows and Linux installs of Modeller now support Python 2.5. (On other platforms you can recompile the Python interface yourself for 2.5.)

• The scaling of rigid body angles in CG and QN optimizations can now be tuned. The restraints file format for rigid bodies has changed accordingly.

• Switch from CHARMM to PDB naming for atoms in nucleic acids. DNA/RNA can now be used by default in alignments by using the a, c, g, t and u one-letter codes.

• Some of the alignment and profile functions have had redundant arguments removed (e.g. profile.scan() no longer takes a write_summary argument, as this is defined by the value of the summary_file argument).

• restraints.spline() now takes Python form, feature and group objects rather than numeric 'spline_select' values.

• The default value of automodel.max_molpdf has been increased from 1e5 to 1e7, since some larger systems can legitimately exceed the smaller value.

• Bugfix: BLK residues written out in PDB files (e.g. in initial models) are now correctly read back in as BLK residues, even if they have the same name as a non-BLK residue type.

• Bugfix: in some cases, dihedral restraints were acting on the wrong atoms after the user called model.patch().

• Bugfix: do not build dihedral restraints (phi/psi, omega) between residues in different chains.

• Bugfix: remember template coordinates, so that automodel.initial_malign3d works correctly and gives reasonable initial models.

• Bugfix: suboptimal alignments are now written to the suboptimal_alignments.out file, not to the log file.

• Bugfix: opening the same gzipped file simultaneously from two separate Modeller processes should now work correctly.

• Bugfix: conflicting omega restraints are now correctly removed from systems containing ligands, even if they are not at the end of the sequence.

• Bugfix: patches that act on two residues (e.g. DISU) can now correctly add or remove atoms that have names longer than 2 characters.