Add support for Keras >=2.2.0

.gitignore

@@ -8,6 +8,7 @@
 __pycache__/*
 .cache/*
 .ipynb_checkpoints
+.pytest_cache
 
 # Project files
 .ropeproject

README.rst

@@ -55,6 +55,7 @@ Table of contents
 News
 ====
 
+* **181201**: DeepCpG 1.0.7 released!
 * **180224**: DeepCpG 1.0.6 released!
 * **171112**: Keras 2 is now the main Keras version (release 1.0.5).
 * **170412**: New `notebook <./examples/notebooks/stats/index.ipynb>`_ on predicting inter-cell statistics!
@@ -213,6 +214,10 @@ Content
 Changelog
 =========
 
+1.0.7
+-----
+* Add support for Keras >=2.2.0.
+
 1.0.6
 -----
 * Add support for Keras 2.1.4 and Tensorflow 1.5.0

deepcpg/__init__.py

@@ -1 +1 @@
-__version__ = '1.0.6'
+__version__ = '1.0.7'

deepcpg/models/utils.py

@@ -6,6 +6,7 @@
 
 from os import path as pt
 
+import keras
 from keras import backend as K
 from keras import models as km
 from keras import layers as kl
@@ -410,6 +411,16 @@ def copy_weights(src_model, dst_model, must_exist=True):
     return copied
 
 
+def is_input_layer(layer):
+    """Test if `layer` is an input layer."""
+    return isinstance(layer, keras.engine.input_layer.InputLayer)
+
+
+def is_output_layer(layer, model):
+    """Test if `layer` is an output layer."""
+    return layer.name in model.output_names
+
+
 class Model(object):
     """Abstract model call.
 
@@ -445,7 +456,7 @@ def _build(self, input, output):
         model = km.Model(input, output, name=self.name)
         if self.scope:
             for layer in model.layers:
-                if layer not in model.input_layers:
+                if not is_input_layer(layer):
                     layer.name = '%s/%s' % (self.scope, layer.name)
         return model
 

docs/source/conf.py

@@ -72,9 +72,9 @@
 # built documents.
 #
 # The short X.Y version.
-version = '1.0.6'
+version = '1.0.7'
 # The full version, including alpha/beta/rc tags.
-release = '1.0.6'
+release = '1.0.7'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

scripts/dcpg_snp.py

@@ -144,9 +144,9 @@ def main(self, name, opts):
 
         # Get DNA layer.
         dna_layer = None
-        for i, name in enumerate(model.input_names):
-            if name == 'dna':
-                dna_layer = model.input_layers[i]
+        for layer in model.layers:
+            if layer.name == 'dna':
+                dna_layer = layer
                 break
         if not dna_layer:
             raise ValueError('The provided model is not a DNA model!')

scripts/dcpg_train.py

@@ -74,6 +74,7 @@
 from deepcpg import data as dat
 from deepcpg import metrics as met
 from deepcpg import models as mod
+from deepcpg.models.utils import is_input_layer, is_output_layer
 from deepcpg.data import hdf, OUTPUT_SEP
 from deepcpg.utils import format_table, make_dir, EPS
 
@@ -86,7 +87,7 @@
 
 
 def remove_outputs(model):
-    while model.layers[-1] in model.output_layers:
+    while is_output_layer(model.layers[-1], model):
         model.layers.pop()
     model.outputs = [model.layers[-1].output]
     model.layers[-1].outbound_nodes = []
@@ -97,7 +98,7 @@ def rename_layers(model, scope=None):
     if not scope:
         scope = model.scope
     for layer in model.layers:
-        if layer in model.input_layers or layer.name.startswith(scope):
+        if is_input_layer(layer) or layer.name.startswith(scope):
             continue
         layer.name = '%s/%s' % (scope, layer.name)
 
@@ -595,7 +596,7 @@ def build_model(self):
             remove_outputs(stem)
 
         outputs = mod.add_output_layers(stem.outputs[0], output_names)
-        model = Model(stem.inputs, outputs, stem.name)
+        model = Model(inputs=stem.inputs, outputs=outputs, name=stem.name)
         return model
 
     def set_trainability(self, model):
@@ -622,17 +623,18 @@ def set_trainability(self, model):
         table['layer'] = []
         table['trainable'] = []
         for layer in model.layers:
-            if layer not in model.input_layers + model.output_layers:
-                if not hasattr(layer, 'trainable'):
-                    continue
-                for regex in not_trainable:
-                    if re.match(regex, layer.name):
-                        layer.trainable = False
-                for regex in trainable:
-                    if re.match(regex, layer.name):
-                        layer.trainable = True
-                table['layer'].append(layer.name)
-                table['trainable'].append(layer.trainable)
+            if is_input_layer(layer) or is_output_layer(layer, model):
+                continue
+            if not hasattr(layer, 'trainable'):
+                continue
+            for regex in not_trainable:
+                if re.match(regex, layer.name):
+                    layer.trainable = False
+            for regex in trainable:
+                if re.match(regex, layer.name):
+                    layer.trainable = True
+            table['layer'].append(layer.name)
+            table['trainable'].append(layer.trainable)
         print('Layer trainability:')
         print(format_table(table))
         print()
@@ -713,9 +715,7 @@ def main(self, name, opts):
         mod.save_model(model, os.path.join(opts.out_dir, 'model.json'))
 
         log.info('Computing output statistics ...')
-        output_names = []
-        for output_layer in model.output_layers:
-            output_names.append(output_layer.name)
+        output_names = model.output_names
 
         output_stats = OrderedDict()
 

setup.py

@@ -9,7 +9,7 @@ def read(fname):
 
 
 setup(name='deepcpg',
-      version='1.0.6',
+      version='1.0.7',
       description='Deep learning for predicting CpG methylation',
       long_description=read('README.rst'),
       author='Christof Angermueller',