automodel() — prepare to build one or more comparative models

automodel(env, alnfile, knowns, sequence, deviation=None, library_schedule=None, csrfile=None, inifile=None, assess_methods=None)

alnfile is required, and usually specifies the name of the PIR file which contains an alignment between knowns (the templates) and sequence (the target sequence).

alnfile can instead be a readable file handle (see modfile.File()) from which the alignment will be read, or an existing alignment object containing knowns and sequence. (Note that this is only supported with a subset of automodel functionality; in particular, it does not work with parallel jobs, automodel.initial_malign3d, or automodel.final_malign3d.)

deviation controls the amount of randomization done by randomize.xyz or randomize.dihedrals; see also automodel.rand_method. (This can also be set after the object is created, by assigning to 'automodel.deviation'. The default is 4Å.)

library_schedule, if given, sets an initial value for automodel.library_schedule

If csrfile is set, restraints are not constructed, but are instead read from the user-supplied file of the same name. See section 2.2.9 for an example.

If inifile is set, an initial model is read from the user-supplied file of the same name. See section 2.2.10 for an example.

assess_methods allows you to request assessment of the generated models (by default, none is done). You can provide a function (or callable), or list of functions, for this purpose, including any of the SOAP potentials (e.g., soap_loop.Scorer(), soap_protein_od.Scorer()), or any of the standard functions provided in the assess module:

• assess.GA341, which uses the GA341 method (see model.assess_ga341()).

• assess.DOPE, which uses the DOPE method (see selection.assess_dope()).

• assess.DOPEHR, which uses the DOPE-HR method (see selection.assess_dopehr()).

• assess.normalized_dope, which uses the normalized DOPE method (see model.assess_normalized_dope()).

(This can also be set after the object is created, by assigning to 'automodel.assess_methods'.) See Section 2.2.3 for an example. Note that only standard models are assessed in this way; if you are also building loop models, see loopmodel.loop.assess_methods.

By default, models are built using heavy atom-only parameters and topology. If you want to use different parameters, read them in before creating the automodel object with Topology.read() and Parameters.read().

See section 2.1 for a general example of using this class.