def main(simmat_path="data/first_110.atg.txt"):
#load similairy matrix
mat = load_simmat(simmat_path)
#convert to distance matrix
dist_mat = sim_to_dist(mat)
#the idx to label function
llf = lambda i: dist_mat.get_label(i)
link_mat = linkage(dist_mat)
disp_linakge_mat(link_mat, llf, len(dist_mat))
grps = get_groups(link_mat, llf)
for g in grps:
print ",".join(g)
def gen_dendrogram(path, out):
"""(str, str) => None
from complex pair-wise distance text file to dendagram
"""
fig = pylab.figure(figsize = (80,40))
mat = load_simmat(path)
print mat.labels
dist_mat = sim_to_dist(mat)
link_mat = linkage(mat)
llf = lambda i: dist_mat.get_label(i)
dendrogram(link_mat, leaf_label_func=llf)
plt.tick_params(axis='both', which='major', labelsize=18)
fig.savefig(out)
