detect_ref.py 文件源码

def download_ref(build, chrom):
    '''Download a chromosome reference file, and remove the header and newlines so we can seek to positions.'''
    dest_filepath = ref_filepath(build, chrom, download=False)
    if os.path.exists(dest_filepath):
        return

    dl_filepath = get_tmp_path('dl-chrom-{}-{}'.format(build['hg'], chrom))
    if not os.path.exists(dl_filepath):
        dl_tmp_filepath = get_tmp_path(dl_filepath)
        url = 'ftp://hgdownload.cse.ucsc.edu/goldenPath/{}/chromosomes/chr{}.fa.gz'.format(build['hg'], chrom)
        wget.download(url=url, out=dl_tmp_filepath)
        print('')
        os.rename(dl_tmp_filepath, dl_filepath)

    tmp_filepath = get_tmp_path(dest_filepath)
    run_script(r'''
    gzip -cd '{dl_filepath}' |
    tail -n +2 |
    tr -d "\n" > '{tmp_filepath}'
    '''.format(dl_filepath=dl_filepath, tmp_filepath=tmp_filepath))
    os.rename(tmp_filepath, dest_filepath)
    print("ref is at", dest_filepath)