def test_rcmatrix_identity():
for order in range(1, 4):
rcmatrix = complement_permmatrix(order)
np.testing.assert_equal(np.eye(pow(4, order)),
np.matmul(rcmatrix, rcmatrix))
def build(self, input_shape):
# from the shape of the one-hot encoding (input_shape),
# we determine the order of the encoding.
self.rcmatrix = K.constant(complement_permmatrix(
int(numpy.log(input_shape[-1]) / numpy.log(4))),
dtype=K.floatx())
super(Complement, self).build(input_shape)
def test_revcomp_rcmatrix(tmpdir):
os.environ['JANGGU_OUTPUT'] = tmpdir.strpath
rcmatrix = complement_permmatrix(1)
np.testing.assert_equal(
rcmatrix,
np.array([[0, 0, 0, 1], [0, 0, 1, 0], [0, 1, 0, 0], [1, 0, 0, 0]]))
rcmatrix = complement_permmatrix(2)
np.testing.assert_equal(
rcmatrix[0], np.array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1]))
np.testing.assert_equal(
rcmatrix[4], np.array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1,
0]))
np.testing.assert_equal(
rcmatrix[8], np.array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0,
0]))
np.testing.assert_equal(
rcmatrix[12],
np.array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0]))
np.testing.assert_equal(
rcmatrix[1], np.array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0,
0]))
np.testing.assert_equal(
rcmatrix[5], np.array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,
0]))
np.testing.assert_equal(
rcmatrix[9], np.array([0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0,
0]))
np.testing.assert_equal(
rcmatrix[13],
np.array([0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0]))
def build(self, input_shape):
with K.name_scope(self.forward_layer.name):
self.forward_layer.build(input_shape)
with K.name_scope(self.revcomp_layer.name):
rcmatrix = K.constant(complement_permmatrix(
int(numpy.log(input_shape[-1]) / numpy.log(4))),
dtype=K.floatx())
kernel = self.forward_layer.kernel[::-1, :, :, :]
kernel = tf.einsum('ij,sdjc->sdic', rcmatrix, kernel)
self.revcomp_layer.kernel = kernel
self.revcomp_layer.bias = self.forward_layer.bias
self.revcomp_layer.use_bias = self.forward_layer.use_bias
self.revcomp_layer.input_spec = self.forward_layer.input_spec
self.built = True
def build(self, input_shape):
with K.name_scope(self.forward_layer._name):
if not self.forward_layer.built:
self.forward_layer.build(input_shape)
with K.name_scope(self.revcomp_layer._name):
rcmatrix = K.constant(complement_permmatrix(
int(numpy.log(input_shape[-1]) / numpy.log(4))),
dtype=K.floatx())
kernel = self.forward_layer.kernel[::-1, :, :, :]
kernel = tf.einsum('ij,sdjc->sdic', rcmatrix, kernel)
self.revcomp_layer.kernel = kernel
self.revcomp_layer.bias = self.forward_layer.bias
self.revcomp_layer.use_bias = self.forward_layer.use_bias
self.revcomp_layer.input_spec = self.forward_layer.input_spec
# these attributes have changed between keras 2.2 and 2.4
# so they need to be set depending on the keras version.
if hasattr(self.forward_layer, '_build_conv_op_input_shape'):
self.revcomp_layer._build_conv_op_input_shape = self.forward_layer._build_conv_op_input_shape
if hasattr(self.forward_layer, '_convolution_op'):
self.revcomp_layer._convolution_op = self.forward_layer._convolution_op
self.built = True
