test_clustering.py 文件源码

def get_subtrees_sklearn(d, bin_chr, bin_position, method="ward", nchrom=1000, distfrac=0.4):

    names = get_names(bin_chr, bin_position)
    #ap = Birch(n_clusters=15)#damping=0.5, max_iter=200, convergence_iter=15, affinity='euclidean') #euclidean precomputed
    ap = KMeans(n_clusters=10)
    assignements = ap.fit_predict(d)#; print assignements[:10]
    c = Counter(assignements); print c.most_common(5)
    subtrees = [[] for i in range(max(assignements)+1)]; print len(subtrees), max(assignements)
    for chrom, i in zip(names, assignements):
        subtrees[i].append(chrom)
    return subtrees