comparison find_str.py @ 23:45f690db0eaf draft

planemo upload for repository https://github.com/fubar2/microsatbed commit d952bc313f408735456747c3d33e09a3170c8f59-dirty

22:b5acf60b903d

"""
Allows all STR or those for a subset of motifs to be written to a bed file
Designed to build some of the microsatellite tracks from https://github.com/arangrhie/T2T-Polish/tree/master/pattern for the VGP.
"""

def getDensity(name, bed, len, winwidth):
    nwin = int(len / winwidth)
    d = [0.0 for x in range(nwin+1)]
    for b in bed:
        nt = b[5]
        bin = int(b[1]/winwidth)
        d[bin] += nt
    dw = [(name,x*winwidth,(x+1)*winwidth,float(d[x])) for x in range(nwin+1) if (x+1)*winwidth <= len]
    return dw

def write_ssrs(args):
    """
    The integers in the call change the minimum repeats for mono-, di-, tri-, tetra-, penta-, hexa-nucleotide repeats

23:45f690db0eaf

"""
Allows all STR or those for a subset of motifs to be written to a bed file
Designed to build some of the microsatellite tracks from https://github.com/arangrhie/T2T-Polish/tree/master/pattern for the VGP.
"""

def getDensity(name, bed, chrlen, winwidth):
    nwin = int(chrlen/winwidth)
    d = [0.0 for x in range(nwin+1)]
    for b in bed:
        nt = b[5]
        bin = int(b[1]/winwidth)
        d[bin] += nt
    dw = [(name,x*winwidth,(x+1)*winwidth,float(d[x])) for x in range(nwin+1) if (x+1)*winwidth <= chrlen]
    return dw

def write_ssrs(args):
    """
    The integers in the call change the minimum repeats for mono-, di-, tri-, tetra-, penta-, hexa-nucleotide repeats