def load_data(self):
print("Building StringDB Graph. It can take a while the first time...")
self.proteinlinks = self.datastore + "/graphs/9606.protein.links.detailed.v11.0.txt"
savefile = self.datastore + "/graphs/stringdb_graph_" + self.graph_type + "_edges.adjlist"
if os.path.isfile(savefile):
self.nx_graph = nx.read_adjlist(savefile)
else:
print(" ensp_to_hugo_map")
ensmap = ensp_to_hugo_map(self.datastore)
print(" reading self.proteinlinks")
edges = pd.read_csv(self.proteinlinks, sep=' ')
selected_edges = edges[self.name_to_edge[self.graph_type]] != 0
edgelist = edges[selected_edges][["protein1",
"protein2"]].values.tolist()
edgelist = [[ensmap[edge[0][5:]], ensmap[edge[1][5:]]]
for edge in edgelist if edge[0][5:] in ensmap.keys()
and edge[1][5:] in ensmap.keys()]
print(" creating OrderedGraph")
self.nx_graph = nx.OrderedGraph(edgelist)
print(" writing graph")
nx.write_adjlist(self.nx_graph, savefile)
# Randomize
if self.randomize:
self.nx_graph = nx.relabel.relabel_nodes(
self.nx_graph, randmap(self.nx_graph.nodes))
print("Graph built !")
def load_data(self):
self.nx_graph = nx.OrderedGraph(
nx.readwrite.gpickle.read_gpickle(
at.get(self.at_hash, datastore=self.datastore)))
# Randomize
if self.randomize:
self.nx_graph = nx.relabel.relabel_nodes(
self.nx_graph, randmap(self.nx_graph.nodes))
def load_data(self):
dir_path = self.datastore #os.path.dirname(os.path.realpath(__file__))
self.location = os.path.join(dir_path, 'graphs/')
pkl_file = os.path.join(self.location, self.filename)
if not os.path.isfile(pkl_file):
self._process_and_pickle(save_name=pkl_file)
self.nx_graph = nx.OrderedGraph(nx.read_gpickle(pkl_file))
# Randomize
if self.randomize:
self.nx_graph = nx.relabel.relabel_nodes(
self.nx_graph, randmap(self.nx_graph.nodes))
def __init__(self, relabel_genes=True, datastore=None, randomize=False):
if datastore is None:
self.datastore = os.path.dirname(os.path.abspath(__file__))
else:
self.datastore = datastore
self.load_data()
self.nx_graph = nx.relabel.relabel_nodes(self.nx_graph, symbol_map(self.nx_graph.nodes))
# Randomize
self.randomize = randomize
if self.randomize:
print("Randomizing the graph")
self.nx_graph = nx.relabel.relabel_nodes(self.nx_graph, randmap(self.nx_graph.nodes))
def load_data(self):
self.benchmark = self.datastore + "/graphs/HumanNet-XN.tsv"
edgelist = pd.read_csv(self.benchmark,
header=None,
sep="\t",
skiprows=1).values[:, :2].tolist()
self.nx_graph = nx.OrderedGraph(edgelist)
# Map nodes from ncbi to hugo names
self.nx_graph = nx.relabel.relabel_nodes(
self.nx_graph,
ncbi_to_hugo_map(self.nx_graph.nodes, datastore=self.datastore))
# Remove nodes which are not covered by the map
for node in list(self.nx_graph.nodes):
if isinstance(node, float):
self.nx_graph.remove_node(node)
# Randomize
if self.randomize:
self.nx_graph = nx.relabel.relabel_nodes(
self.nx_graph, randmap(self.nx_graph.nodes))
"train_size": train_size,
"seed": seed,
}
print(experiment)
X_train, X_test, y_train, y_test = sklearn.model_selection.train_test_split(features,
target.to_numpy(),
stratify=target.to_numpy(),
train_size=train_size,
test_size=test_size,
shuffle=True,
random_state=seed
)
if gene_graph.randomize:
print("Randomizing the graph")
gene_graph.nx_graph = nx.relabel.relabel_nodes(gene_graph.nx_graph, randmap(gene_graph.nx_graph.nodes))
if num_genes == 'all':
if 'MLP' in model_name:
x_train = X_train.copy()
x_test = X_test.copy()
adj = None
else:
neighbors = gene_graph.nx_graph
intersection_nodes = np.intersect1d(X_train.columns, neighbors.nodes)
x_train = X_train[list(intersection_nodes)].copy()
x_test = X_test[list(intersection_nodes)].copy()
toremove = set(neighbors.nodes)
toremove = toremove.difference(intersection_nodes)
neighbors.remove_nodes_from(toremove)
"seed": seed,
}
print(experiment)
X_train, X_test, y_train, y_test = sklearn.model_selection.train_test_split(
features,
target.to_numpy(),
stratify=target.to_numpy(),
train_size=train_size,
test_size=test_size,
shuffle=True,
random_state=seed)
if gene_graph.randomize:
print("Randomizing the graph")
gene_graph.nx_graph = nx.relabel.relabel_nodes(
gene_graph.nx_graph, randmap(gene_graph.nx_graph.nodes))
if num_genes == 'all':
if 'MLP' in model_name:
x_train = X_train.copy()
x_test = X_test.copy()
adj = None
else:
neighbors = gene_graph.nx_graph
intersection_nodes = np.intersect1d(X_train.columns,
neighbors.nodes)
x_train = X_train[list(intersection_nodes)].copy()
x_test = X_test[list(intersection_nodes)].copy()
toremove = set(neighbors.nodes)
toremove = toremove.difference(intersection_nodes)