def parse(hmmin): self.scores = hmmer.parse_scores(hmmin)
prev = hits[0]
prev_name,_ = prev
buf = []
maxScores = []
for hit in hits:
name,score = hit
if name==prev_name:
buf.append(score)
else:
maxScores.append(max(buf))
buf = [score]
prev_name = name
return maxScores
folder = "/Users/mortonyt/Documents/MiamiBio/workspace"
toxin_scores = hmmer.parse_scores("%s/boa_scores.out"%folder)
#modifier_scores = hmmer.parse_scores("%s/modifier.out"%folder)
#immunity_scores = hmmer.parse_scores("%s/immunity.out"%folder)
#regulator_scores = hmmer.parse_scores("%s/regulator.out"%folder)
#transport_scores = hmmer.parse_scores("%s/transport.out"%folder)
bagel_scores = hmmer.parse_scores("%s/bagel_toxin.out"%folder)
#all_scores = numpy.array(toxin_scores+modifier_scores+immunity_scores+regulator_scores+transport_scores)
all_scores = toxin_scores
print all_scores[:10]
print "Number of all scores",len(all_scores)
print "Number of bagel scores",len(bagel_scores)
gmm = GaussianMixtureModel(all_scores)
#params = gmm.expectation_maximization(1000)
g = mixture.GMM(n_components=2)
model = g.fit(all_scores)
x = numpy.linspace(min(all_scores),max(all_scores),1000)