blast_on_cluster.py

import sys
from datetime import datetime
from math import ceil
import shlex
import subprocess

cluster_template = """#!/bin/bash
#

#$ -N blast_on_cluster
#$ -cwd
#$ -j y
#$ -S /bin/bash
#$ -o OUT_$JOB_NAME.$JOB_ID
#$ -e ERR_$JOB_NAME.$JOB_ID

module load biotools/ncbi-blast-2.3.0+
date
hostname
blastp -outfmt 6 -evalue 0.001 -query ${q} -db ${db} -out ${o} -num_threads {t}
date
"""


def check_line(line):
    """
    Checks if a line is a blastp command

    :param line: string to check
    :return: True if the line is a blastp command
    """
    return line.startswith("blastp -outfmt")


def write_template(filename=None):
    if filename is None:
        filename = "blast_on_cluster" + str(ceil(datetime.utcnow().timestamp())) + ".sh"

    with open(filename, "w") as t:
        print(cluster_template, file=t)

    return filename


def execute_command_on_cluster(submission_script, cmd, threads=1):
    """
    Parses blastp command reported by OrthoFinder and submits an equivalent job to the cluster

    :param submission_script: path to the submission script
    :param cmd: command reported by OrthoFinder
    :param threads: number of desired thread for each BLAST
    :return:
    """
    cmd_args = shlex.split(cmd)

    query, db, output = cmd_args[6], cmd_args[8], cmd_args[10]

    qsub_cmd = ['qsub',
                '-pe', 'cores', str(threads),
                '-v', 'q=%s,db=%s,out=%s,t=%d' % (query, db, output, threads),
                submission_script]

    print(qsub_cmd)
    # subprocess.call(qsub_cmd)


def run():
    """
    Main loop, will get STDIN, parse the line and submit the job using qsub
    """
    print("Writing template")
    submission_script = write_template()

    print("Waiting for command to execute...")

    for line in sys.stdin():
        if check_line(line):
            print("\n\tCommand found! Executing")
            execute_command_on_cluster(submission_script, line.strip())

if __name__ == "main":
    run()