sam-discard-dups.py 文件源码

def filter_sam_dups(filename,ids_to_discard):
    """
    Reads a SAM file, returns a set() of duplicated reads
    (reads which are listed more than once)
    """
    try:
        sam=file(filename,'r')

        for linenum,line in enumerate(sam):
            err = "input error in '%s' line %d: " % (filename, linenum+1)
            line = line.strip()

            if line[:1]=='@':
                print line
                continue

            flds = line.split('\t')
            read_id = flds[0]

            if not (read_id in ids_to_discard):
                print line

    except IOError as e:
        if e.errno == errno.EPIPE:
            sys.exit(0) # exit silently

        # TODO: this is a cop-out, hard to tell what's the exact error
        #       and give informative,useful error message to the user.
        sys.exit("I/O error: %s" % (str(e)))