def test_state_loading(self):
a = GeneralizedPairState()
a.load(self.inputY)
X = a.toJSON()
Y = self.inputY
X['emission'].sort()
Y['emission'].sort()
X['durations'].sort()
Y['durations'].sort()
self.assertDictEqual(X, Y, "Loading and dumping to JSON does not " + \
" work: \n" + str(X) + " != \n" + str(Y))
def __init__(self, *args, **_):
GeneralizedPairState.__init__(self, *args)
self.dp = config.preparers[config.preparer_index][0](constants.window_size)
self.clf_fname = 'data/clf/{}{}{}.clf'.format(
PairClassifier.get_name(),
config.preparers[config.preparer_index][2],
constants.window_size,
)
self.clf = self._get_classifier()
self.annotations, self.ann_x, self.ann_y = None, None, None
self.emission_table = None
def test_state_loading(self):
a = GeneralizedPairState()
a.load(self.inputY)
X = a.toJSON()
Y = self.inputY
X['emission'].sort()
Y['emission'].sort()
X['durations'].sort()
Y['durations'].sort()
self.assertDictEqual(X, Y, "Loading and dumping to JSON does not " + \
" work: \n" + str(X) + " != \n" + str(Y))
def setUp(self):
self.inputY = {
"__name__": "GeneralizedPairState",
"name": "name",
"startprob": 1.0,
"emission": [(("0", "0"), 1.0), (("00", ""), 1.0)],
"endprob": 0.5,
"durations": [((1, 1), 0.5), ((2, 0), 0.5)]
}
self.HMM = dict()
for mathType in self.mathTypes:
hmmInit = {
"__name__": "GeneralizedPairHMM",
"states": [],
"transitions": [
{"from": "one", "to": "one", "prob": mathType(0.3)},
{"from": "one", "to": "two", "prob": mathType(0.6)},
{"from": "one", "to": "three", "prob": mathType(0.1)},
{"from": "two", "to": "one", "prob": mathType(0.5)},
{"from": "two", "to": "two", "prob": mathType(0.4)},
{"from": "two", "to": "three", "prob": mathType(0.1)},
{"from": "three", "to": "one", "prob": mathType(0.2)},
{"from": "three", "to": "two", "prob": mathType(0.3)},
{"from": "three", "to": "three", "prob": mathType(0.5)},
]
}
one = GeneralizedPairState(mathType)
one.load({
"__name__": "GeneralizedPairState",
"name": "one",
"startprob": 0.2,
"endprob": 1.0,
"emission": [(("0", "0"), 1.0), (("00", "00"), 1.0)],
"durations": [((1, 1), 0.5), ((2, 2), 0.5)]
})
two = GeneralizedPairState(mathType)
two.load({
"__name__": "GeneralizedPairState",
"name": "two",
"startprob": 0.8,
"endprob": 1.0,
"emission": [(("0", ""), 1.0), (("00", "0"), 1.0)],
"durations": [((1, 0), 0.5), ((2, 1), 0.5)]
})
three = GeneralizedPairState(mathType)
three.load({
"__name__": "GeneralizedPairState",
"name": "three",
"startprob": 0.8,
"endprob": 1.0,
"emission": [(("", "0"), 1.0), (("0", "00"), 1.0)],
"durations": [((0, 1), 0.5), ((1, 2), 0.5)]
})
hmmInit["states"] = [one, two, three]
self.HMM[mathType] = GeneralizedPairHMM(mathType)
self.HMM[mathType].load(hmmInit)
def __init__(self, *args, **_):
GeneralizedPairState.__init__(self, *args)
self.dp = config.preparers[config.preparer_index][0](
constants.window_size)
self.clf_fname = 'data/clf/{}{}{}.clf'.format(
PairClassifier.get_name(),
config.preparers[config.preparer_index][2],
constants.window_size,
)
self.clf = self._get_classifier()
self.annotations, self.ann_x, self.ann_y = None, None, None
self.emission_table = None
def test_state(self):
for numType in self.mathTypes:
state = GeneralizedPairState(numType)
state.load(self.inputY)
#test duration
X = list(state.durationGenerator())
Y = [((1, 1), numType(0.5)), ((2, 0), numType(0.5))]
self.assertEqual(X, Y, "HMM.durationGenerator() does not work: " + \
str(X) + " != " + str(Y))
#test emission
Y = numType(1.0)
X = state.emission("000", 1, 1, "000", 2, 1)
self.assertAlmostEqual(X, Y, delta=1e-7,
msg="HMM.emission(\"000\", 1, 1) does not " \
+ "work: " + str(X) + " != " + str(Y))
Y = numType(1.0)
X = state.emission("000", 1, 2, "000000", 2, 0)
self.assertAlmostEqual(X, Y, delta=1e-7,
msg="HMM.emission(\"000\", 1, 2) does not " \
+ "work: " + str(X) + " != " + str(Y))
def test_state(self):
for numType in self.mathTypes:
state = GeneralizedPairState(numType)
state.load(self.inputY)
#test duration
X = list(state.durationGenerator())
Y = [((1, 1), numType(0.5)), ((2, 0), numType(0.5))]
self.assertEqual(X, Y, "HMM.durationGenerator() does not work: " + \
str(X) + " != " + str(Y))
#test emission
Y = numType(1.0)
X = state.emission("000", 1, 1, "000", 2, 1)
self.assertAlmostEqual(X, Y, delta=1e-7,
msg="HMM.emission(\"000\", 1, 1) does not " \
+ "work: " + str(X) + " != " + str(Y))
Y = numType(1.0)
X = state.emission("000", 1, 2, "000000", 2, 0)
self.assertAlmostEqual(X, Y, delta=1e-7,
msg="HMM.emission(\"000\", 1, 2) does not " \
+ "work: " + str(X) + " != " + str(Y))
def setUp(self):
self.inputY = {
"__name__": "GeneralizedPairState",
"name": "name",
"startprob": 1.0,
"emission": [(("0", "0"), 1.0), (("00", ""), 1.0)],
"endprob": 0.5,
"durations": [((1, 1), 0.5), ((2, 0), 0.5)]
}
self.HMM = dict()
for mathType in self.mathTypes:
hmmInit = {
"__name__":
"GeneralizedPairHMM",
"states": [],
"transitions": [
{
"from": "one",
"to": "one",
"prob": mathType(0.3)
},
{
"from": "one",
"to": "two",
"prob": mathType(0.6)
},
{
"from": "one",
"to": "three",
"prob": mathType(0.1)
},
{
"from": "two",
"to": "one",
"prob": mathType(0.5)
},
{
"from": "two",
"to": "two",
"prob": mathType(0.4)
},
{
"from": "two",
"to": "three",
"prob": mathType(0.1)
},
{
"from": "three",
"to": "one",
"prob": mathType(0.2)
},
{
"from": "three",
"to": "two",
"prob": mathType(0.3)
},
{
"from": "three",
"to": "three",
"prob": mathType(0.5)
},
]
}
one = GeneralizedPairState(mathType)
one.load({
"__name__": "GeneralizedPairState",
"name": "one",
"startprob": 0.2,
"endprob": 1.0,
"emission": [(("0", "0"), 1.0), (("00", "00"), 1.0)],
"durations": [((1, 1), 0.5), ((2, 2), 0.5)]
})
two = GeneralizedPairState(mathType)
two.load({
"__name__": "GeneralizedPairState",
"name": "two",
"startprob": 0.8,
"endprob": 1.0,
"emission": [(("0", ""), 1.0), (("00", "0"), 1.0)],
"durations": [((1, 0), 0.5), ((2, 1), 0.5)]
})
three = GeneralizedPairState(mathType)
three.load({
"__name__": "GeneralizedPairState",
"name": "three",
"startprob": 0.8,
"endprob": 1.0,
"emission": [(("", "0"), 1.0), (("0", "00"), 1.0)],
"durations": [((0, 1), 0.5), ((1, 2), 0.5)]
})
hmmInit["states"] = [one, two, three]
self.HMM[mathType] = GeneralizedPairHMM(mathType)
self.HMM[mathType].load(hmmInit)
def expand(self, params=None):
if params == None:
raise "Houston, we have a problem"
consensuses = params['consensus']
prefix_t = self.stateName + '_{}_'
prefix_i = 0
states = list()
transitions = list()
init = self.stateName + '_Init'
end = self.stateName + '_End'
total = 0.0
for consensus in consensuses:
prefix = prefix_t.format(prefix_i)
prefix_i += 1
model = self.factory.getHMM(consensus)
probability = self.consensusDistribution[consensus]
total += probability
json = model.toJSON()
new_init = prefix + 'Init'
new_end = prefix + 'End'
st = map(deepcopy, model.states)
for i in range(len(st)):
st[i].stateName = prefix + st[i].stateName
tr = map(
lambda x: {
'from': prefix + x['from'],
'to': prefix + x['to'],
'prob': x['prob'],
},
json['transitions'],
)
st, tr, new_init, new_end = double_track_hmm(
st, tr, new_init, new_end, self.mathType
)
transitions.extend(tr)
states.extend(st)
transitions.extend([
{
'from': init,
'to': new_init,
'prob': probability,
},
{
'from': new_end,
'to': end,
'prob': 1.0,
},
])
#transitions.append({
# 'from': init,
# 'to': init,
# 'prob': 1.0 - total,
#})
template = {
'__name__': 'GeneralizedPairState',
'name': init,
'startprob': 1.0,
'endprob': 0.0,
'emission': [(('', ''), self.mathType(1.0))],
'durations': [((0, 0), self.mathType(1.0))],
}
st = GeneralizedPairState(self.mathType)
st.load(template)
states.append(st)
template['name'] = end
template['startprob'] = 0.0
template['endprob'] = 1.0
st = GeneralizedPairState(self.mathType)
st.load(template)
states.append(st)
return states, transitions, init, end
