def test_log_add(self):
cases = ((1, 2), (0.5, 0.5), (50, 0.7))
error = 0.0000000001
for x, y in cases:
expected = log(x + y)
result = findcpgsites.log_add(log(x), log(y))
difference = abs(expected - result)
self.assertTrue(difference < error)
result = findcpgsites.log_add(log(y), log(x))
difference = abs(expected - result)
self.assertTrue(difference < error)
def test_log_add(self):
cases = (
(1, 2),
(0.5, 0.5),
(50, 0.7)
)
error = 0.0000000001
for x, y in cases:
expected = log(x + y)
result = findcpgsites.log_add(log(x), log(y))
difference = abs(expected - result)
self.assertTrue(difference < error)
result = findcpgsites.log_add(log(y), log(x))
difference = abs(expected - result)
self.assertTrue(difference < error)
def test_calc_backward_probabilities(self):
sequence = 'CGCA'
expected_prob_arr = np.zeros((8, 4))
expected_prob_arr[0, 3] = expected_prob_arr[4, 3] = log(1.0 / 8)
expected_prob_arr[1, 2] = findcpgsites.log_add(
expected_prob_arr[0, 3] + LOG_PROBS[1, 0],
expected_prob_arr[4, 3] + LOG_PROBS[1, 4])
expected_prob_arr[5, 2] = findcpgsites.log_add(
expected_prob_arr[0, 3] + LOG_PROBS[5, 0],
expected_prob_arr[4, 3] + LOG_PROBS[5, 4])
expected_prob_arr[2, 1] = findcpgsites.log_add(
expected_prob_arr[1, 2] + LOG_PROBS[2, 1],
expected_prob_arr[5, 2] + LOG_PROBS[2, 5])
expected_prob_arr[6, 1] = findcpgsites.log_add(
expected_prob_arr[1, 2] + LOG_PROBS[6, 1],
expected_prob_arr[5, 2] + LOG_PROBS[6, 5])
expected_prob_arr[1, 0] = findcpgsites.log_add(
expected_prob_arr[2, 1] + LOG_PROBS[1, 2],
expected_prob_arr[6, 1] + LOG_PROBS[1, 6])
expected_prob_arr[5, 0] = findcpgsites.log_add(
expected_prob_arr[2, 1] + LOG_PROBS[5, 2],
expected_prob_arr[6, 1] + LOG_PROBS[5, 6])
expected_final_prob = findcpgsites.log_add(
expected_prob_arr[1, 0] + log(1.0 / 8),
expected_prob_arr[5, 0] + log(1.0 / 8))
result_prob_arr, result_final_prob = (
findcpgsites.calc_backward_probabilities(sequence, LOG_PROBS))
self.assertTrue(np.array_equal(result_prob_arr, expected_prob_arr))
self.assertEqual(result_final_prob, expected_final_prob)
def test_calc_forward_probabilities(self):
sequence = 'CGCA'
expected_prob_arr = np.zeros((8,4))
expected_prob_arr[1,0] = expected_prob_arr[5,0] = log(1.0 / 8)
expected_prob_arr[2,1] = findcpgsites.log_add(
expected_prob_arr[1,0] + LOG_PROBS[1,2],
expected_prob_arr[5,0] + LOG_PROBS[5,2]
)
expected_prob_arr[6,1] = findcpgsites.log_add(
expected_prob_arr[1,0] + LOG_PROBS[1,6],
expected_prob_arr[5,0] + LOG_PROBS[5,6]
)
expected_prob_arr[1,2] = findcpgsites.log_add(
expected_prob_arr[2,1] + LOG_PROBS[2,1],
expected_prob_arr[6,1] + LOG_PROBS[6,1]
)
expected_prob_arr[5,2] = findcpgsites.log_add(
expected_prob_arr[2,1] + LOG_PROBS[2,5],
expected_prob_arr[6,1] + LOG_PROBS[6,5]
)
expected_prob_arr[0,3] = findcpgsites.log_add(
expected_prob_arr[1,2] + LOG_PROBS[1,0],
expected_prob_arr[5,2] + LOG_PROBS[5,0]
)
expected_prob_arr[4,3] = findcpgsites.log_add(
expected_prob_arr[1,2] + LOG_PROBS[1,4],
expected_prob_arr[5,2] + LOG_PROBS[5,4]
)
expected_final_prob = findcpgsites.log_add(
expected_prob_arr[0,3] + log(1.0 / 8),
expected_prob_arr[4,3] + log(1.0 / 8)
)
result_prob_arr, result_final_prob = (
findcpgsites.calc_forward_probabilities(
sequence, LOG_PROBS)
)
self.assertTrue(
np.array_equal(result_prob_arr, expected_prob_arr))
self.assertEqual(result_final_prob, expected_final_prob)
