download_genes.py 文件源码

def run(argv):
    gencode_filepath = get_generated_path('sites/genes/gencode-{}.gtf.gz'.format(genes_version))
    genes_filepath = common_filepaths['genes']

    if not os.path.exists(genes_filepath):
        print('genes-{}.bed will be stored at {!r}'.format(genes_version, genes_filepath))
        if not os.path.exists(gencode_filepath):
            make_basedir(gencode_filepath)
            wget.download(
                url="ftp://ftp.sanger.ac.uk/pub/gencode/Gencode_human/release_25/GRCh37_mapping/gencode.v25lift37.annotation.gtf.gz",
                out=gencode_filepath
            )
            print('')
        genes = get_all_genes(gencode_filepath)
        genes = dedup_ensg(genes)
        genes = dedup_symbol(genes)

        make_basedir(genes_filepath)
        with open(genes_filepath, 'w') as f:
            writer = csv.DictWriter(f, delimiter='\t', fieldnames='chrom start end symbol ensg'.split(), lineterminator='\n')
            writer.writerows(genes)

    else:
        print("gencode is at {!r}".format(genes_filepath))