utils.py 文件源码

def iter_by_qname(in_genome_bam, in_trimmed_bam):
    # Iterate through multiple BAMs by qname simultaneously
    # Assume the trimmed-read-bam has every qname in the genome bam, in the same order.

    genome_bam_iter = itertools.groupby(in_genome_bam, key=lambda read: read.qname)

    if in_trimmed_bam is None:
        trimmed_bam_iter = iter(())
    else:
        trimmed_bam_iter = itertools.groupby(in_trimmed_bam, key=lambda read: read.qname)

    for (genome_qname, genome_reads), trimmed_tuple in itertools.izip_longest(genome_bam_iter,
                                                                              trimmed_bam_iter):
        trimmed_qname, trimmed_reads = trimmed_tuple or (None, [])
        genome_reads = list(genome_reads)
        trimmed_reads = list(trimmed_reads)

        assert (in_trimmed_bam is None) or trimmed_qname == genome_qname
        yield (genome_qname, genome_reads, trimmed_reads)