Functions.py 文件源码

def construct_pssm(cds, length=90, out_path="", prob=None):
    """
    Construct Position Specific Scoring Matrices with log-likelihood values
    length: size of analyzed region from start, in bp  (sequences that are not this size are discarded)
    prob : a dict of bases with a priori expected probabilities  
    """
    cds = cds[0]
    if not prob:
        prob = {"a":0.25, "t":0.25, "g":0.25, "c":0.25}
    m = {"a":[0]*length, "t":[0]*length, "g":[0]*length, "c":[0]*length}
    tot_gene = 0.0
    for gene in cds:
        if len(cds[gene]) >= length:
            tot_gene += 1
            for i in range(length):
                m[cds[gene][i]][i] += 1
    for k in m:
        m[k] = [log((v/tot_gene)/prob[k]) for v in m[k]]
    if out_path:
        h = open(out_path, "w")
        h.write(","+",".join([str(i) for i in range(1,length+1)])+"\n")
        for b in ["a", "t", "g", "c"]:
            h.write(b+","+",".join(["%.2f" % v for v in m[b]])+"\n")
        h.close()
    return m