prophyle_analyze.py 文件源码

def open_asg(in_fn, in_format):

    # try to detect kraken and histo formats automatically
    if in_format is None:
        if '.' in in_fn:
            f_ext = in_fn.split('.')[-1]
            if f_ext in IN_FMTS:
                in_format = f_ext
        else:
            with open(in_fn, 'rb') as f:
                f_start = f.read(2)
                if f_start == b'C\t' or f_start == b'U\t':
                    in_format = 'kraken'
                elif f_start == b'#O':
                    in_format = 'histo'

    if in_format == 'sam':
        in_f = pysam.AlignmentFile(in_fn, "r")
    elif in_format == 'bam':
        in_f = pysam.AlignmentFile(in_fn, "rb")
    elif in_format == 'cram':
        in_f = pysam.AlignmentFile(in_fn, "rc")
    elif in_format == 'uncompressed_bam':
        in_f = pysam.AlignmentFile(in_fn, "ru")
    elif in_format == 'kraken':
        in_f = open(in_fn, 'r')
    # no need to load assignments if input is a histogram -> go to load_histo
    elif in_format == 'histo':
        in_f = None
    # let pysam assess the format
    else:
        in_f = pysam.AlignmentFile(in_fn)
    return in_f, in_format