utils.py 文件源码

def split_genes_by_genomes(genes, genomes):
    """ Returns a list of lists [genome1, genome2, ...]
    where genome1 = [gene1,gene2,...].
    Args:
      genes - list of Gene tuples
      genomes - list of genome names, e.g. ['hg19', 'mm10']
    """
    assert len(genomes) > 0

    if len(genomes) == 1:
        return [genes]

    d = collections.defaultdict(list)
    for gene in genes:
        genome = get_genome_from_str(gene.id, genomes)
        d[genome].append(gene)

    genes_per_genome = []
    for genome in genomes:
        genes_per_genome.append(d[genome])
    return genes_per_genome