Changes since release 7v7

• TOP is no longer used as the scripting language. MODELLER now reads Python 2.3 scripts instead. (TOP scripts can still be used, but will not take advantage of all new features.) See Appendix C for more information.

• Incorporation of the GA341 (model.assess_ga341()) and DOPE (selection.assess_dope()) methods for model assessment.

• `Automatic' filename generation (when filenames were set to 'default') is no longer supported; use the modfile.default() function instead.

• Memory is used much more efficiently, and various limits evident in older releases (e.g., errors concerning MAXRES, MAXCNS and MAXBND) have been removed. Consequently, you can no longer assign to the MAXRES variable; this is not necessary now.

• New alignment.salign() alignment command, using structural information, and based on the COMPARER program.

• Various fixes to profile.build(), and a new command profile.scan(), for building and using sequence profiles.

• Generated models can be docked to electron microscopy density maps, using the new density.grid_search() command.

• Superpositions are now improved by using local substructures. This local refinement can be disabled (thus getting the behavior of old MODELLER versions) by turning off the refine_local option to selection.superpose().

• MODELLER will now read compressed files in the .gz and .bz2 formats (as well as the existing .Z format) provided the gzip and bzip2 programs are available.

• Fixes for various problems reading alignment files with very long lines, or with Mac, Unix or DOS-style line endings.

• Bug fix for alignment.compare_structures(); disulfide bridges were not being properly considered in the comparison.

• Charges for all topologies are now assigned using a `nearest neighbor' technique; see topology.make().

• sequence_db.search() (formerly SEQUENCE_SEARCH) now takes a 1-sequence alignment as input, rather than reading in that alignment itself.