CrossMapper.py

import urllib.request
import os
from Modules import CrossMap


class CrossMapper:

    """

    Class to download chain_files for chrossmapping hg38 or mm10 to older assembly versions.
    Utilizes CrossMap.py. see wiki for more information.

    @author: Sebastian Beyvers
    @contact: sebastian.beyvers@med.uni-giessen.de


    """

    def __init__(self, org, wd, out, is_dir):

        # Constructor for CrossMapper class
        # input_parameter: org    = input organism
        #                  wd     = working directory
        #                  out    = path to output-file -> Parameter
        #                  is_dir = boolean if wd is data_dir or just working directory

        # Get path to tempfile / outputfile and chain-file

        if is_dir:
            self.infile = os.path.join(wd + "/temp/" + org + ".gtf")
        else:
            self.infile = os.path.join(wd+"/data/temp/"+org+".gtf")
        self.outfile = os.path.join(out)
        self.chainfile = self.get_chain_file(org, wd, is_dir)

        # Execute Crossmap for gff/gtf files

        (mapTree, targetChromSizes, sourceChromSizes) = CrossMap.read_chain_file(self.chainfile)

        # Map results and save output to self.outfile

        CrossMap.crossmap_gff_file(mapTree, self.infile, self.outfile)

    def get_chain_file(self, org, wd, is_data_dir):

        # Defines the Chain files for different conversions.
        # input_parameter: org         = organism
        #                  wd          = working directory
        #                  is_data_dir = is wd data_dir or not

        # return_value: Link to chain-file for conversion.
        # Custom chain-files and chain-files for more organism can be specified in this section

        if org == "hg19":
            if is_data_dir:
                file_link = os.path.join(wd+"temp/hg38tohg19.over.chain.gz")
            else:
                file_link = os.path.join(wd + "/data/temp/hg38tohg19.over.chain.gz" )
            urllib.request.urlretrieve("http://hgdownload.soe.ucsc.edu/goldenPath/hg38/"
                                       "liftOver/hg38ToHg19.over.chain.gz",
                                       file_link)
            return file_link

        elif org == "mm9":
            if is_data_dir:
                file_link = os.path.join(wd+"temp/mm10ToMm9.over.chain.gz")
            else:
                file_link = os.path.join(wd + "/data/temp/hg38tohg19.over.chain.gz" )
            urllib.request.urlretrieve("http://hgdownload.soe.ucsc.edu/goldenPath/mm10/"
                                       "liftOver/mm10ToMm9.over.chain.gz",
                                       file_link)
            return file_link