pdb.py 文件源码

def download_biological_assemblies(pdb_id, outdir):
    """Downloads biological assembly file from:
    `ftp://ftp.wwpdb.org/pub/pdb/data/biounit/coordinates/divided/`

    Args:
        outdir (str): Output directory of the decompressed assembly

    """

    # TODO: not tested yet
    if not op.exists(outdir):
        raise ValueError('{}: output directory does not exist'.format(outdir))

    folder = pdb_id[1:3]
    server = 'ftp://ftp.wwpdb.org/pub/pdb/data/biounit/coordinates/divided/{}/'.format(folder)
    html_folder = urlopen(server).readlines()
    for line in html_folder:
        if pdb_id in str(line).strip():
            file_name = '%s' % (pdb_id + str(line).strip().split(pdb_id)[1].split('\r\n')[0])
            outfile_name = file_name.replace('.', '_')
            outfile_name = outfile_name.replace('_gz', '.pdb')
            f = urlopen(op.join(server, file_name))
            decompressed_data = zlib.decompress(f.read(), 16 + zlib.MAX_WBITS)
            with open(op.join(outdir, outfile_name), 'wb') as f:
                f.write(decompressed_data)
                f.close()
            log.debug('{}: downloaded biological assembly')
            return op.join(outdir, outfile_name)