2_1_mcflashfold_mask_generator.py

#!/usr/bin/env python
# -*- coding: UTF-8 -*-

#
# This utility is part of the DecoyDB (working title) project
#
# Author: Stephen Leong Koan
#
# Copyright (C) 2013 Université de Montréal
#

"""
Command-line utility to generate an MC-FlashFold mask to make sure the part where the mature are either unpaired or opening/closing bracket
"""

import os
import argparse
import cPickle

if __name__ == '__main__':
    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)

    parser.add_argument('--digested_data',
                        action="store",
                        required=True,
                        dest="digested_data",
                        help="The pickle file generated by 1_organize_precursor_mature.py")
    parser.add_argument('--mcfold_cmd',
                        action="store",
                        required=True,
                        dest="mcfold_cmd",
                        help=("The command line template. Put '{seq}', '{mask}', '{name}', '{accession}', "
                              "where you want the scriot to plug the sequence, the mask, the name of the precursor "
                              "and/or the accession of the precursor respectively"))

    ns = parser.parse_args()

    digested_data = ns.digested_data
    mcfold_cmd = ns.mcfold_cmd

    list_digested_data = None

    with open(digested_data, 'rb') as dd:
        list_digested_data = cPickle.load(dd)

    for hairpin_dict in list_digested_data:
        hairpin_name = hairpin_dict['name']
        hairpin_acc = hairpin_dict['accession']
        hairpin_seq = hairpin_dict['sequence']
        mature_5p_range = []
        mature_3p_range = []
        for mature in hairpin_dict['matures']:
            if mature.get("alternative_name", mature["name"]).endswith("5p"):
                start = hairpin_seq.find(mature["sequence"])
                mature_5p_range = range(start, start + len(mature["sequence"]))
            else:
                start = hairpin_seq.find(mature["sequence"])
                mature_3p_range = range(start, start + len(mature["sequence"]))

        mask = ""

        for i in xrange(len(hairpin_seq)):
            if i in mature_5p_range:
                mask += "p"
            elif i in mature_3p_range:
                mask += "q"
            else:
                mask += "x"

        cmd = mcfold_cmd.format(seq=hairpin_seq,
                                mask=mask,
                                accession=hairpin_acc,
                                name=hairpin_name)
        print cmd
#         print ("python git/various-codes/DecoyDB/2_2_mcflashfold_structure_filter.py "
#                "--hairpin_fasta reproduction_projet_naim/2D/hairpin/{acc} "
#                "--mature_fasta reproduction_projet_naim/2D/mature/{acc} "
#                "--mcfold_output reproduction_projet_naim/MC-FlashFold_part1/2d/{acc} > reproduction_projet_naim/MC-FlashFold_part1/2dclean/{acc}").format(acc=hairpin_acc)