Create script to generate ncbitaxon obo graph and add it to mira dock…

…er image

docker/Dockerfile

@@ -1,3 +1,16 @@
+# Create an initial docker image to generate the graph and transfer it to the
+# second docker image
+# We do this to avoid involving additional imports in the second docker image
+FROM python:3.10-slim AS graph-builder
+
+WORKDIR /graphs
+RUN apt-get update && apt-get install -y git
+RUN pip install pyobo networkx obonet
+
+# Copy and run the script to generate the pickled graph
+COPY generate_graph.py /graphs/generate_graph.py
+RUN python generate_graph.py
+
 FROM ubuntu:focal
 
 WORKDIR /sw
@@ -42,5 +55,7 @@ RUN python -m pip install --upgrade pip && \
 # Copy the example json for reconstructing the ode semantics
 RUN wget -O /sw/sir_flux_span.json https://raw.githubusercontent.com/gyorilab/mira/main/tests/sir_flux_span.json
 
+RUN mkdir -p /graphs
+COPY --from=graph-builder /graphs/relabeled_obo_graph.pkl /graphs/relabeled_obo_graph.pkl
 COPY startup.sh startup.sh
 ENTRYPOINT ["/bin/bash", "/sw/startup.sh"]

docker/generate_graph.py

@@ -0,0 +1,26 @@
+from pyobo import get_version
+from pyobo.getters import _ensure_ontology_path
+from pathlib import Path
+from obonet import read_obo
+import networkx
+import pickle
+
+def download_convert_ncbitaxon_obo_to_graph():
+ resource_prefix = "ncbitaxon"
+ version = get_version(resource_prefix)
+
+ # Checks to see if the pickled ncbitaxon obo graph exists in the container
+ cached_relabeled_obo_graph_path = Path("/graphs/relabeled_obo_graph.pkl")
+ if not cached_relabeled_obo_graph_path.exists():
+ _, obo_path = _ensure_ontology_path(resource_prefix, force=False,
+ version=version)
+ obo_graph = read_obo(obo_path)
+ relabeled_graph = networkx.relabel_nodes(obo_graph,
+ lambda node_index:
+ node_index.lower())
+ with open(cached_relabeled_obo_graph_path,
+ 'wb') as relabeled_graph_file:
+ pickle.dump(relabeled_graph, relabeled_graph_file)
+
+if __name__ == "__main__":
+ download_convert_ncbitaxon_obo_to_graph()