Opossum.py 文件源码

def TrimCigar(cigar) :

    clippings = [1 for cigartype, cigarlength in cigar if cigartype in [4,5]]

    if not clippings :
        return cigar, 0, None

    start = 0
    end = 0

    # soft and hard clippings can either be in the beginning or end of a read
    for cigartype, cigarlength in cigar[:] :

        if cigartype == 4 : # soft clipping
            start = cigarlength
            cigar.remove((cigartype, cigarlength))

        elif cigartype == 5 : # hard clipping
            cigar.remove((cigartype, cigarlength))

        else :
            break

    for cigartype, cigarlength in reversed(cigar[:]) :

        if cigartype == 4 : # soft clipping
            end -= cigarlength
            cigar.remove((cigartype, cigarlength))

        elif cigartype == 5 : # hard clipping
            cigar.remove((cigartype, cigarlength))

        else :
            break

    return cigar, start, end if not end == 0 else None




# Give as input the quality string, md from Tags field, cigar, and mincut and maxcut,
# and ipos (starting index of current read in terms of quality sequence)
# and adjustbase (only needed if two reads
# are being merged, and second read starts later than first read).
# Md tells whether there are any base-changes lying closer or equal to mincut from
# read start edge, or at least maxcut away from same edge. These base-changes will be 
# discarded, that is, their corresponding base
# quality will be set to 0 (or ! in ascii format).
# Returns updated quality string.