def get_vntr_matcher_hmm(self, read_length):
"""Try to load trained HMM for this VNTR
If there was no trained HMM, it will build one and store it for later usage
"""
logging.info('Using read length %s' % read_length)
copies = self.get_copies_for_hmm(read_length)
base_name = str(
self.reference_vntr.id) + '_' + str(read_length) + '.json'
stored_hmm_file = settings.TRAINED_HMMS_DIR + base_name
if settings.USE_TRAINED_HMMS and os.path.isfile(stored_hmm_file):
model = Model()
model = model.from_json(stored_hmm_file)
return model
flanking_region_size = read_length
vntr_matcher = self.build_vntr_matcher_hmm(copies,
flanking_region_size)
json_str = vntr_matcher.to_json()
with open(stored_hmm_file, 'w') as outfile:
outfile.write(json_str)
return vntr_matcher
def build_reference_repeat_finder_hmm(patterns, copies=1):
pattern = patterns[0]
model = Model(name="HMM Model")
insert_distribution = DiscreteDistribution({
'A': 0.25,
'C': 0.25,
'G': 0.25,
'T': 0.25
})
last_end = None
start_random_matches = State(insert_distribution,
name='start_random_matches')
end_random_matches = State(insert_distribution, name='end_random_matches')
model.add_states([start_random_matches, end_random_matches])
for repeat in range(copies):
insert_states = []
match_states = []
delete_states = []
for i in range(len(pattern) + 1):
insert_states.append(
State(insert_distribution, name='I%s_%s' % (i, repeat)))
for i in range(len(pattern)):
distribution_map = dict({
'A': 0.01,
'C': 0.01,
'G': 0.01,
'T': 0.01
})
distribution_map[pattern[i]] = 0.97
match_states.append(
State(DiscreteDistribution(distribution_map),
name='M%s_%s' % (str(i + 1), repeat)))
for i in range(len(pattern)):
delete_states.append(
State(None, name='D%s_%s' % (str(i + 1), repeat)))
unit_start = State(None, name='unit_start_%s' % repeat)
unit_end = State(None, name='unit_end_%s' % repeat)
model.add_states(insert_states + match_states + delete_states +
[unit_start, unit_end])
last = len(delete_states) - 1
if repeat > 0:
model.add_transition(last_end, unit_start, 0.5)
else:
model.add_transition(model.start, unit_start, 0.5)
model.add_transition(model.start, start_random_matches, 0.5)
model.add_transition(start_random_matches, unit_start, 0.5)
model.add_transition(start_random_matches, start_random_matches,
0.5)
model.add_transition(unit_end, end_random_matches, 0.5)
if repeat == copies - 1:
model.add_transition(unit_end, model.end, 0.5)
model.add_transition(end_random_matches, end_random_matches, 0.5)
model.add_transition(end_random_matches, model.end, 0.5)
model.add_transition(unit_start, match_states[0], 0.98)
model.add_transition(unit_start, delete_states[0], 0.01)
model.add_transition(unit_start, insert_states[0], 0.01)
model.add_transition(insert_states[0], insert_states[0], 0.01)
model.add_transition(insert_states[0], delete_states[0], 0.01)
model.add_transition(insert_states[0], match_states[0], 0.98)
model.add_transition(delete_states[last], unit_end, 0.99)
model.add_transition(delete_states[last], insert_states[last + 1],
0.01)
model.add_transition(match_states[last], unit_end, 0.99)
model.add_transition(match_states[last], insert_states[last + 1], 0.01)
model.add_transition(insert_states[last + 1], insert_states[last + 1],
0.01)
model.add_transition(insert_states[last + 1], unit_end, 0.99)
for i in range(0, len(pattern)):
model.add_transition(match_states[i], insert_states[i + 1], 0.01)
model.add_transition(delete_states[i], insert_states[i + 1], 0.01)
model.add_transition(insert_states[i + 1], insert_states[i + 1],
0.01)
if i < len(pattern) - 1:
model.add_transition(insert_states[i + 1], match_states[i + 1],
0.98)
model.add_transition(insert_states[i + 1],
delete_states[i + 1], 0.01)
model.add_transition(match_states[i], match_states[i + 1],
0.98)
model.add_transition(match_states[i], delete_states[i + 1],
0.01)
model.add_transition(delete_states[i], delete_states[i + 1],
0.01)
model.add_transition(delete_states[i], match_states[i + 1],
0.98)
last_end = unit_end
model.bake()
if len(patterns) > 1:
# model.fit(patterns, algorithm='baum-welch', transition_pseudocount=1, use_pseudocount=True)
fit_patterns = [pattern * copies for pattern in patterns]
model.fit(fit_patterns,
algorithm='viterbi',
transition_pseudocount=1,
use_pseudocount=True)
return model
def get_constant_number_of_repeats_matcher_hmm(patterns, copies):
model = Model(name="Repeating Pattern Matcher HMM Model")
transitions, emissions = build_profile_hmm_for_repeats(
patterns, settings.MAX_ERROR_RATE)
matches = [m for m in emissions.keys() if m.startswith('M')]
last_end = None
for repeat in range(copies):
insert_states = []
match_states = []
delete_states = []
for i in range(len(matches) + 1):
insert_distribution = DiscreteDistribution(emissions['I%s' % i])
insert_states.append(
State(insert_distribution, name='I%s_%s' % (i, repeat)))
for i in range(1, len(matches) + 1):
match_distribution = DiscreteDistribution(emissions['M%s' % i])
match_states.append(
State(match_distribution, name='M%s_%s' % (str(i), repeat)))
for i in range(1, len(matches) + 1):
delete_states.append(State(None, name='D%s_%s' % (str(i), repeat)))
unit_start = State(None, name='unit_start_%s' % repeat)
unit_end = State(None, name='unit_end_%s' % repeat)
model.add_states(insert_states + match_states + delete_states +
[unit_start, unit_end])
n = len(delete_states) - 1
if repeat > 0:
model.add_transition(last_end, unit_start, 1)
else:
model.add_transition(model.start, unit_start, 1)
if repeat == copies - 1:
model.add_transition(unit_end, model.end, 1)
model.add_transition(unit_start, match_states[0],
transitions['unit_start']['M1'])
model.add_transition(unit_start, delete_states[0],
transitions['unit_start']['D1'])
model.add_transition(unit_start, insert_states[0],
transitions['unit_start']['I0'])
model.add_transition(insert_states[0], insert_states[0],
transitions['I0']['I0'])
model.add_transition(insert_states[0], delete_states[0],
transitions['I0']['D1'])
model.add_transition(insert_states[0], match_states[0],
transitions['I0']['M1'])
model.add_transition(delete_states[n], unit_end,
transitions['D%s' % (n + 1)]['unit_end'])
model.add_transition(delete_states[n], insert_states[n + 1],
transitions['D%s' % (n + 1)]['I%s' % (n + 1)])
model.add_transition(match_states[n], unit_end,
transitions['M%s' % (n + 1)]['unit_end'])
model.add_transition(match_states[n], insert_states[n + 1],
transitions['M%s' % (n + 1)]['I%s' % (n + 1)])
model.add_transition(insert_states[n + 1], insert_states[n + 1],
transitions['I%s' % (n + 1)]['I%s' % (n + 1)])
model.add_transition(insert_states[n + 1], unit_end,
transitions['I%s' % (n + 1)]['unit_end'])
for i in range(1, len(matches) + 1):
model.add_transition(match_states[i - 1], insert_states[i],
transitions['M%s' % i]['I%s' % i])
model.add_transition(delete_states[i - 1], insert_states[i],
transitions['D%s' % i]['I%s' % i])
model.add_transition(insert_states[i], insert_states[i],
transitions['I%s' % i]['I%s' % i])
if i < len(matches):
model.add_transition(insert_states[i], match_states[i],
transitions['I%s' % i]['M%s' % (i + 1)])
model.add_transition(insert_states[i], delete_states[i],
transitions['I%s' % i]['D%s' % (i + 1)])
model.add_transition(match_states[i - 1], match_states[i],
transitions['M%s' % i]['M%s' % (i + 1)])
model.add_transition(match_states[i - 1], delete_states[i],
transitions['M%s' % i]['D%s' % (i + 1)])
model.add_transition(delete_states[i - 1], match_states[i],
transitions['D%s' % i]['M%s' % (i + 1)])
model.add_transition(delete_states[i - 1], delete_states[i],
transitions['D%s' % i]['D%s' % (i + 1)])
last_end = unit_end
model.bake(merge=None)
return model
def get_suffix_matcher_hmm(pattern):
model = Model(name="Suffix Matcher HMM Model")
insert_distribution = DiscreteDistribution({
'A': 0.25,
'C': 0.25,
'G': 0.25,
'T': 0.25
})
insert_states = []
match_states = []
delete_states = []
hmm_name = 'suffix'
for i in range(len(pattern) + 1):
insert_states.append(
State(insert_distribution, name='I%s_%s' % (i, hmm_name)))
for i in range(len(pattern)):
distribution_map = dict({'A': 0.01, 'C': 0.01, 'G': 0.01, 'T': 0.01})
distribution_map[pattern[i]] = 0.97
match_states.append(
State(DiscreteDistribution(distribution_map),
name='M%s_%s' % (str(i + 1), hmm_name)))
for i in range(len(pattern)):
delete_states.append(
State(None, name='D%s_%s' % (str(i + 1), hmm_name)))
unit_start = State(None, name='suffix_start_%s' % hmm_name)
unit_end = State(None, name='suffix_end_%s' % hmm_name)
model.add_states(insert_states + match_states + delete_states +
[unit_start, unit_end])
last = len(delete_states) - 1
model.add_transition(model.start, unit_start, 1)
model.add_transition(unit_end, model.end, 1)
model.add_transition(unit_start, delete_states[0], 0.01)
model.add_transition(unit_start, insert_states[0], 0.01)
for i in range(len(pattern)):
model.add_transition(unit_start, match_states[i], 0.98 / len(pattern))
model.add_transition(insert_states[0], insert_states[0], 0.01)
model.add_transition(insert_states[0], delete_states[0], 0.01)
model.add_transition(insert_states[0], match_states[0], 0.98)
model.add_transition(delete_states[last], unit_end, 0.99)
model.add_transition(delete_states[last], insert_states[last + 1], 0.01)
model.add_transition(match_states[last], unit_end, 0.99)
model.add_transition(match_states[last], insert_states[last + 1], 0.01)
model.add_transition(insert_states[last + 1], insert_states[last + 1],
0.01)
model.add_transition(insert_states[last + 1], unit_end, 0.99)
for i in range(0, len(pattern)):
model.add_transition(match_states[i], insert_states[i + 1], 0.01)
model.add_transition(delete_states[i], insert_states[i + 1], 0.01)
model.add_transition(insert_states[i + 1], insert_states[i + 1], 0.01)
if i < len(pattern) - 1:
model.add_transition(insert_states[i + 1], match_states[i + 1],
0.98)
model.add_transition(insert_states[i + 1], delete_states[i + 1],
0.01)
model.add_transition(match_states[i], match_states[i + 1], 0.98)
model.add_transition(match_states[i], delete_states[i + 1], 0.01)
model.add_transition(delete_states[i], delete_states[i + 1], 0.01)
model.add_transition(delete_states[i], match_states[i + 1], 0.98)
model.bake(merge=None)
return model
infinite model, with no extra work! This change is passed on to all the
algorithms.
'''
from pomegranate import *
from pomegranate import HiddenMarkovModel as Model
import itertools as it
import numpy as np
# Define the states
s1 = State(NormalDistribution(5, 2), name="S1")
s2 = State(NormalDistribution(15, 2), name="S2")
s3 = State(NormalDistribution(25, 2), name="S3 ")
# Define the transitions
model = Model("infinite")
model.add_transition(model.start, s1, 0.7)
model.add_transition(model.start, s2, 0.2)
model.add_transition(model.start, s3, 0.1)
model.add_transition(s1, s1, 0.6)
model.add_transition(s1, s2, 0.1)
model.add_transition(s1, s3, 0.3)
model.add_transition(s2, s1, 0.4)
model.add_transition(s2, s2, 0.4)
model.add_transition(s2, s3, 0.2)
model.add_transition(s3, s1, 0.05)
model.add_transition(s3, s2, 0.15)
model.add_transition(s3, s3, 0.8)
model.bake()
sequence = [4.8, 5.6, 24.1, 25.8, 14.3, 26.5, 15.9, 5.5, 5.1]
# Contact: Jacob Schreiber # [email protected] """ Example rainy-sunny HMM using yahmm. Example drawn from the wikipedia HMM article: http://en.wikipedia.org/wiki/Hidden_Markov_model describing what Bob likes to do on rainy or sunny days. """ from pomegranate import * from pomegranate import HiddenMarkovModel as Model import random import math random.seed(0) model = Model(name="Rainy-Sunny") # Emission probabilities rainy = State(DiscreteDistribution({ 'walk': 0.1, 'shop': 0.4, 'clean': 0.5 }), name='Rainy') sunny = State(DiscreteDistribution({ 'walk': 0.6, 'shop': 0.3, 'clean': 0.1 }), name='Sunny')
#!/usr/bin/env python2.7 # example.py: Yet Another Hidden Markov Model library # Contact: Jacob Schreiber ( [email protected] ) """ A simple example highlighting how to build a model using states, add transitions, and then run the algorithms, including showing how training on a sequence improves the probability of the sequence. """ import random from pomegranate import * from pomegranate import HiddenMarkovModel as Model random.seed(0) model = Model(name="ExampleModel") distribution = UniformDistribution(0.0, 1.0) state = State(distribution, name="uniform") state2 = State(NormalDistribution(0, 2), name="normal") silent = State(None, name="silent") model.add_state(state) model.add_state(state2) model.add_transition(state, state, 0.4) model.add_transition(state, state2, 0.4) model.add_transition(state2, state2, 0.4) model.add_transition(state2, state, 0.4) model.add_transition(model.start, state, 0.5) model.add_transition(model.start, state2, 0.5) model.add_transition(state, model.end, 0.2) model.add_transition(state2, model.end, 0.2)