gregorMakeConfig

#! /usr/bin/env python3

# Vivek Rai
# vivekrai@umich.edu
# The Parker Lab
#
# Updated: Oct 16, 2018
# GPLv3


__description__ = """Generate a GREGOR config file using supplied options."""


import os
import sys
import glob
import argparse
from sys import stderr


def create_index_bed(bedfilename, annotfiles, annotdir):
    if os.path.exists(bedfilename):
        print("File already exists, so will not overwite.", file=stderr)
        return

    with open(bedfilename, "w") as f:
        if annotfiles is not None:
            for bed_file in annotfiles:
                f.write("{path}\n".format(path=os.path.abspath(bed_file)))
        elif annotdir is not None:
            for bed_file in glob.glob(os.path.join(annotdir + "*.bed")):
                f.write("{path}\n".format(path=os.path.abspath(bed_file)))
        else:
            print("No suitable config found for creating `indexbed.txt`", file=stderr)


def getopts():
    parser = argparse.ArgumentParser(description=__description__)
    group = parser.add_mutually_exclusive_group()

    parser.add_argument(
        "--outfile",
        type=str,
        default="gregor.conf",
        help="""Name of output .conf file""",
    )

    # User must either supply "annotfiles" or a "directory with annotation files".
    # Use argparse's group to implement this.
    group.add_argument(
        "--annotfiles",
        nargs="+",
        type=str,
        help="""File with paths of annotation files to calculate enrichment in.""",
    )
    group.add_argument(
        "--annotdir",
        type=str,
        help="""The directory with (unzipped) bed files if --annotfiles are not
        specified. Will take all .bed files in this directory""",
    )

    parser.add_argument(
        "--indexbed",
        type=str,
        default="indexbed.txt",
        help="""Name of file to be created with paths of annotfiles""",
    )
    parser.add_argument(
        "--snpfile",
        type=str,
        help="""chr:pos formatted SNP files. Filename *MUST* not contain `_`""",
    )
    parser.add_argument(
        "--refdir",
        type=str,
        default="/lab/data/sw/GREGOR/1.4.0/0.7",
        help="""GREGOR reference. (default: /lab/data/sw/GREGOR/1.4.0/0.7)""",
    )
    parser.add_argument(
        "--population", type=str, default="EUR", help="""Population (default: EUR)"""
    )
    parser.add_argument(
        "-r2",
        "--gregorR2Threshold",
        default="0.8",
        help="""Minimum LD r2 for proxy SNPs (default: 0.8)""",
    )
    parser.add_argument(
        "--ldwindow",
        type=int,
        default=1000000,
        help="""Minimum LD r2 for proxy SNPs (default: 0.8)""",
    )
    parser.add_argument(
        "-d", "--outdir", type=str, help="""Name of the output directory"""
    )
    parser.add_argument(
        "--neighbor", type=int, default=500, help="""Min neighbors for GREGOR"""
    )
    parser.add_argument(
        "--issorted",
        type=str,
        default="true",
        help="""true or false if bed files are sorted""",
    )
    parser.add_argument("--topnbed", type=int, default=2, help="""Top n bed""")
    parser.add_argument(
        "--cores",
        type=str,
        default="6",
        help="""Number of cores for each GREGOR job. (default: 6)""",
    )
    parser.add_argument(
        "--batchtype",
        type=str,
        default="local",
        help="""Batch job submission type (default: local)""",
    )
    return parser


if __name__ == "__main__":
    parser = getopts()
    args = parser.parse_args()

    args.indexbed = os.path.abspath(args.indexbed)
    create_index_bed(args.indexbed, args.annotfiles, args.annotdir)

    with open(args.outfile, "w") as f:
        f.write("INDEX_SNP_FILE = {} \n".format(args.snpfile))
        f.write("BED_FILE_INDEX = {} \n".format(args.indexbed))
        f.write("REF_DIR = {} \n".format(args.refdir))
        f.write("POPULATION = {} \n".format(args.population))
        f.write("R2THRESHOLD = {} \n".format(args.gregorR2Threshold))
        f.write("LDWINDOWSIZE = {} \n".format(args.ldwindow))
        f.write("OUT_DIR = {} \n".format(args.outdir))
        f.write("MIN_NEIGHBOR_NUM = {} \n".format(args.neighbor))
        f.write("BEDFILE_IS_SORTED = {} \n".format(args.issorted))
        f.write("TOPNBEDFILES = {} \n".format(args.topnbed))
        f.write("JOBNUMBER = {} \n".format(args.cores))
        f.write("BATCHTYPE = {} \n".format(args.batchtype))