def __init__(self,
n_components=10,
n_clusters=3,
selected_embedding='autoencoder',
level_dims_in=LEVEL_DIMS_IN,
level_dims_out=LEVEL_DIMS_OUT,
dropout=DROPOUT,
epochs=EPOCHS,
_clustering=CLUSTERING,
_embedding=EMBEDDING,
**kwargs):
"""
"""
self.level_dims_in = level_dims_in
self.level_dims_out = level_dims_out
self.dropout = dropout
self.epochs = epochs
_embedding['autoencoder'] = DeepBase(
dropout=self.dropout,
epochs=self.epochs,
level_dims_in=self.level_dims_in,
level_dims_out=self.level_dims_out,
n_components=n_components)
_clustering['autoencoder'] = DeepBase(
dropout=self.dropout,
epochs=self.epochs,
level_dims_in=self.level_dims_in,
level_dims_out=self.level_dims_out,
n_components=n_clusters)
GranatumClustering.__init__(self,
selected_embedding=selected_embedding,
n_components=n_components,
n_clusters=n_clusters,
_embedding=_embedding,
_clustering=_clustering,
**kwargs)
self._plotting_embedding['autoencoder'] = DeepBase(
dropout=self.dropout,
epochs=self.epochs,
level_dims_in=self.level_dims_in,
level_dims_out=self.level_dims_out,
n_components=2)
self._find_best_cluster['autoencoder'] = {
'method': _silhouette_analysis,
'range': FIND_BEST_CLUSTER_RANGE
}
def main():
inputs = json.load(stdin)
granatum_clustering = GranatumClustering(**inputs)
matrix = inputs['assay']['matrix']
matrix = np.array(matrix).T
sample_ids = inputs['assay']['sampleIds']
results = granatum_clustering.fit(matrix=matrix,
sample_ids=sample_ids,
jsonify=True)
stdout.write(json.dumps(results))
def test_5_default():
"""
test all plotting embbeding algorithm
"""
granatum = GranatumClustering(
n_components=5)
granatum.fit(TEST_DATASET)
for embbed in PLOTTING_EMBEDDING:
print('#### plotting embbeding used tested: {0}'.format(embbed))
plots = granatum.plot(embedding=embbed)
assert('embedding_for_plotting' in plots)
assert('plot_figure_html' in plots)
assert('plot_figure_png' in plots)
def test_4_default():
"""
test all clustering algorithm with optimum cluster selection
"""
for clustering in CLUSTERING:
print('#### clustering algo tested: {0}'.format(clustering))
granatum = GranatumClustering(
n_components=5,
selected_clustering=clustering)
results = granatum.fit(TEST_DATASET)
print('number of clusters found: {0}'.format(results['n_clusters']))
assert('clusters' in results)
assert('n_clusters' in results)
assert('n_components' in results)
assert('embedding' in results)
assert('clustering_algorithm' in results)
def test_3_default():
"""
test all clustering_algorithm
"""
for clustering in CLUSTERING:
print('#### clustering algo tested: {0}'.format(clustering))
granatum = GranatumClustering(
n_components=5,
selected_plotting_embedding='PCA',
selected_clustering=clustering,
find_best_number_of_cluster=False)
results = granatum.fit(TEST_DATASET)
assert('clusters' in results)
assert('n_clusters' in results)
assert('n_components' in results)
assert('embedding' in results)
assert('clustering_algorithm' in results)
def test_2_default():
"""
test all algorithm for embedding
"""
for embedd in EMBEDDING:
print('#### embedding tested: {0}'.format(embedd))
granatum = GranatumClustering(
n_components=5,
find_best_number_of_cluster=False,
selected_embedding=embedd)
results = granatum.fit(TEST_DATASET)
assert('clusters' in results)
assert('n_clusters' in results)
assert('n_components' in results)
assert('embedding' in results)
assert('clustering_algorithm' in results)
def test_0_default():
"""
test default parameter on the default test dataset and the plotting
"""
granatum = GranatumClustering(n_components=5, selected_clustering='DBSCAN')
results = granatum.fit(TEST_DATASET)
assert('clusters' in results)
assert('n_clusters' in results)
assert('n_components' in results)
assert('embedding' in results)
assert('clustering_algorithm' in results)
plots = granatum.plot(plot_figure_png=True,
plot_figure_html=True,
embedding='PCA',
jsonify=False)
assert('embedding_for_plotting' in plots)
assert('plot_figure_html' in plots)
assert('plot_figure_png' in plots)
def main():
""" """
from sklearn.datasets import make_blobs
test_dataset, ref_array = make_blobs(n_samples=300,
n_features=200,
centers=8)
metadata = {
i: {
'dummy': 'dummy:{0}'.format(i)
}
for i in range(len(test_dataset))
}
sample_ids = list(range(len(test_dataset)))
granatum = GranatumClustering(selected_embedding='PCA',
selected_clustering='WARD')
results = granatum.fit(matrix=test_dataset,
metadata=metadata,
sample_ids=sample_ids)
assert ('clusters' in results)
assert ('n_clusters' in results)
assert ('n_components' in results)
assert ('embedding' in results)
assert ('clustering_algorithm' in results)
plots = granatum.plot(figsize=(600, 600),
embedding='PCA',
plot_figure_png=True,
plot_figure_html=True,
jsonify=False)
assert ('plot_figure_html' in plots)
assert ('plot_figure_png' in plots)