view get_data/kegg_glycan/getkcfKEGG.py @ 0:89592faa2875 draft

Uploaded

__author__ = 'cbarnett'
__license__ = "MIT"
__version = "0.3"


def helper_get_G(lineentry):
    if "G" in lineentry.upper():
        splitline = lineentry.split()
        for item in splitline:
            if "G" in item:
                return item
    return None


def get_kcf_from_kegg(inputstream):
    """
    :param inputstream: input stream containing gl entries
    :return: list of kcf output and list of db entry output or empty list
    """
    import urllib2

    uri = 'http://rest.kegg.jp/get/'
    if inputstream is None or inputstream == []:
        raise IOError("empty input stream")
    dbresponses = []
    kcfresponses = []
    for line in inputstream:
        glentry = helper_get_G(line)
        if glentry is not None:
            try:
                dbresponse = urllib2.urlopen(uri + glentry).read()
                kcfresponse = urllib2.urlopen(uri + glentry + "/kcf").read()
            except Exception as e:
                raise urllib2.HTTPError(e.url, e.code, e.msg, e.hdrs, e.fp)
            dbresponses.append(dbresponse)
            kcfresponses.append(kcfresponse)
    return kcfresponses, dbresponses


if __name__ == "__main__":
    from optparse import OptionParser

    usage = "usage: python %prog [options]\n"
    parser = OptionParser(usage=usage)
    parser.add_option("-i", action="store", type="string", dest="i", default="input",
                      help="single or double column text file containing GL entries")
    parser.add_option("-k", action="store", type="string", dest="k", default="kcf.output",
                      help="kcf output file name")
    parser.add_option("-d", action="store", type="string", dest="d", default="db.output",
                      help="KEGG db entry in text format output file name")
    (options, args) = parser.parse_args()
    try:
        instream = file(options.i, 'r')
    except Exception as e:
        raise IOError(e, "the input file specified does not exist. Use -h flag for help")
    kcf, db = get_kcf_from_kegg(instream)
    try:
        kcfout = file(options.k, 'w')
        dbout = file(options.d, 'w')
    except Exception as e:
        raise IOError(e, "cannot open output files. -h flag for help")

    kcfout.write("".join(kcf))
    dbout.write("".join(db))
    kcfout.close()
    dbout.close()