Run NCBI BLAST with an easy-to-use Pythonic API.

Running NCBI BLAST manually is of course not rocket science, but this module provides several benefits over doing so:

• Automatically runs a BLAST process for each CPU on the system; achieves far better throughput than the -num_threads option
• Provides an iterator API that emits native Python objects for each BLAST result as they're produced, rather than at the end
• Result and Hit objects obviate the need for manually parsing results; all values represented by their native Python types (e.g. Hit.evalue is a float, etc)

Here's a simple example with comments hilighting some relevant features

import pyblast

with open('data.fasta') as f:
    # Use the pyblast.blastx() iterator function
    for r in pyblast.blastx(f, db='/path/to/swissprot'):
        msg = 'query {} has {} hits'.format(r.query_id, len(r.hits))
        if r.hits:
            # Use Hit.evalue as a float for comparison
            min_evalue = sorted([h.evalue for h in r.hits])[0]
            msg += '; minimum evalue {:f}'.format(min_evalue)

        print msg

This will produce output like the following

query M00181:167:000000000-A4VBV:1:1101:11880:1874 1:N:0:6 has 6 hits; minimum evalue 0.310000
query M00181:167:000000000-A4VBV:1:1101:17067:1875 1:N:0:6 has 14 hits; minimum evalue 0.200000
query M00181:167:000000000-A4VBV:1:1101:15039:1878 1:N:0:6 has 4 hits; minimum evalue 4.400000
query M00181:167:000000000-A4VBV:1:1101:17090:1895 1:N:0:6 has 6 hits; minimum evalue 1.700000
query M00181:167:000000000-A4VBV:1:1101:15843:1907 1:N:0:6 has 2 hits; minimum evalue 1.800000

blastn(input_file, *args, **kwargs)

Iterator to process the contents of the FASTA input_file using blastn; yields Result objects.

The *args and **kwargs arguments control how blastn is invoked. The former are passed as options without values, while the latter are passed as options with values. For example, blastn(some_file, 'ungapped', db='foo/bar') will run blastn with the -ungapped -db foo/bar options.

In addition, the following keyword arguments are handled specially and are not passed on to BLAST:

• pb_num_processes: number of BLAST processes to spawn; default is sysconf(SC_NPROCESSORS_ONLN)
• pb_fields: iterable of field names to retrieve for each hit; default is DEFAULT_HIT_FIELDS. The list of valid field names (and their meanings) can be found in the *** Formatting options section of blastn -help.

blastp(input_file, *args, **kwargs)

See documentation for blastn.

blastx(input_file, *args, **kwargs)

See documentation for blastn.

Result

The result of BLAST processing a single query sequence. The set of attributes on this object are:

• id: identifier for the query sequence; can be None
• description: textual description of the query sequence; can be None
• hits: array of Hit objects

Hit

A single sequence hit in a Result object.

The attributes of this object are the names of the fields requested of BLAST. For example, if blastn was run with pb_fields=['qseqid', ...] then one could access the qseqid value of the Hit object h like so: h.qseqid. Fields referenced that were not requested of BLAST have a None value.

In addition, BLAST fields are converted to their native Python types. For example, evalue fields are automatically converted to floating point values.

DEFAULT_HIT_FIELDS

The default fields returned for each Hit object.

VERSION

The version of pyblast that's being used. This can be used to more easily than feature detection to determine what features of the module are available.