.. _conf:

The ModPipe configuration file
==============================

The configuration file, specified by the :option:`--conf_file` command line
argument to many ModPipe programs, provides file locations, such as the
location of the template sequence and profile files, and the location of
ModPipe output files. The configuration file also provides a number of run
parameters, such as whether template sequences will be clustered before
building models. The variables in the configuration file are
described here.

Environment variables can be used in the configuration file using familiar
Unix syntax - e.g. ``$FOO`` is replaced with the value of the environment
variable ``FOO``.

See also :ref:`databases` for information on the databases pointed to by
this file, and information on setting up these databases to run ModPipe
if you are not in the Sali lab.

See also a `sample configuration file <modpipe.conf>`_.

:envvar:`TMPDIR`
   ModPipe will create a new directory here for every sequence it processes
   and will use this as scratch space for all calculations. This directory
   should be local to the machine running ModPipe in order to reduce
   network traffic.

:envvar:`DATDIR`
   This is the base directory in which the
   :ref:`ModPipe filesystem <filesystem>` will be created.

:envvar:`TEMPLATESEQDB`
   The name of the file of template sequences to use in searches for matches.
   Unless you have some special need and know what you are doing you should
   use a binary (HDF5) database file; for example, 
   :file:`/netapp/sali/ModPipe/database/PDB95/db/pdb_95.hdf5`.

:envvar:`XPRF_LIST`
   Name of file containing list of template profile (:file:`.prf`) files --
   one for each template in the :envvar:`TEMPLATESEQDB` database file.

:envvar:`XPRF_PSSMDB`
   Name of the file containing position-specific scoring matrix data for each
   template sequence.

:envvar:`PDB_REPOSITORY`
   The name of the directory containing PDB files.

:envvar:`NRSEQDB`
   The name of the file of non-redundant sequences to use for construction
   of profiles by Modeller's :meth:`Profile.build`. Unless you have some
   special need and know what you are doing you should use a binary (HDF5)
   database file; for example,
   :file:`/netapp/database/uniprot/sequences/uniprot90.hdf5`.

:envvar:`NCBISEQDB`
   The name of the file of non-redundant sequences to use for construction of 
   profiles by PSI-BLAST.

:envvar:`NRDBTAG`
   A short-name for the non-redundant sequence database that ModPipe will use
   as part of the name of profile files (multiple sequence alignments)
   constructed using that database. Usually it will be :file:`uniprot90`.

:envvar:`PRFUPDATE`
   If this flag is set to *ON*, irrespective of the existence of a profile for
   the target sequence, a new profile will be calculated. If 
   :envvar:`PRFUPDATE` is set to *OFF*, it will calculate a new
   target-sequence profile only if one does not already exist.

:envvar:`NUMMODELS`
   Number of models to calculate for each alignment.

:envvar:`CLUSTERALI`
   When using multiple :ref:`fold assignment methods <fold_assignment_methods>`,
   since they are used independently they will typically find a number of
   templates in common. By setting this variable to *ON*, redundant and highly
   similar hits in the alignment of all templates found will be removed.