Adding Directionality

CONTRIBUTING.md

@@ -189,22 +189,13 @@ First pull the image `<username>/local-neighborhood` from Docker Hub.
 Then build the Docker image using the `Dockerfile` that was completed in Step 2.
 
 Modify Generate Inputs:
-
-Purpose: Update the references and expected files for the new algorithm.
-
-Instructions:
-1. Append the algorithm name to the existing algorithms list.
-2. Include a key-value pair in the algo_exp_file dictionary that links the specific algorithm to its expected network file.
-3. Store the expected network file at /test/GenerateInputs/expected using the naming convention: {algorithm name}-{network file name}-expected.txt.
+1. Include a key-value pair in the algo_exp_file dictionary that links the specific algorithm to its expected network file.
+2. Create or obtain the expected network file and save it to `/test/GenerateInputs/expected`. Name it as: `{algorithm name}-{network file name}-expected.txt`.
 
 Modify Parse Outputs:
-
-Purpose: Handle and verify the outputs of the new algorithm.
-
-Instructions:
-1. Save the raw-pathway output (from the run function in your wrapper) to /test/ParseOutputs/input. Name it as  {algorithm name}-raw-pathway.txt.
-2. Place the expected parsed output in the test/ParseOutputs/expected directory, following the naming pattern: {algorithm name}-pathway-expected.txt.
-3. Add the new algorithm's name to the algorithms list.
+1. Obtain the raw-pathway output (ex. from the run function in your wrapper by running the Snakemake workflow) and save it to `/test/ParseOutputs/input`. Name it as  `{algorithm name}-raw-pathway.txt`.
+2. Create or obtain the expected universal output and save it to `test/ParseOutputs/expected` directory. Name it as: `{algorithm name}-pathway-expected.txt`.
+3. Add the new algorithm's name to the algorithms list in `test/ParseOutputs/test_parse_outputs.py`.
 
 ### Step 6: Work with SPRAS maintainers to revise the pull request
 Step 0 previously described how to create a `local-neighborhood` branch and create a pull request.
@@ -223,7 +214,7 @@ The pull request will be closed so that the `master` branch of the fork stays sy
 1. Import the new class in `src/runner.py` so the wrapper functions can be accessed
 1. Document the usage of the Docker wrapper and the assumptions made when implementing the wrapper
 1. Add example usage for the new algorithm and its parameters to the template config file
-1. Write test functions and provide example input data in a new test subdirectory `test/<algorithm>`
+1. Write test functions and provide example input data in a new test subdirectory `test/<algorithm>`. Provide example data and algorithm/expected files names to lists or dicts in `test/GenerateInputs` and `test/ParseOutputs`.
 1. Extend `.github/workflows/test-spras.yml` to pull and build the new Docker image
 
 When adding new algorithms, there are many other considerations that are not relevant with the simple Local Neighborhood example.

spras/analysis/graphspace.py

@@ -8,10 +8,8 @@
 from graphspace_python.graphs.classes.gsgraph import GSGraph
 
 
-# remove all the the directed = bool in the function names
 def write_json(graph_file,out_graph,out_style) -> None:
 
-	print("graph file", graph_file)
 	# get GS Graph
 	graph_name = os.path.basename(out_graph) # name is the prefix specified.
 	G = get_gs_graph(graph_file, graph_name)
@@ -40,8 +38,14 @@ def post_graph(G:GSGraph,username:str,password:str) -> None:
 	return
 
 def get_gs_graph(graph_file:str,graph_name:str) -> GSGraph:
+	"""
+	Creates a GraphSpace graph using the networkx graph or digraph and directionality returned from load_graph
+
+	If the graph is empty, that likely means the graph was mized in directionality
+	- this will return an emty GraphSpace graph
+	"""
 	# read file as networkx graph
-	# returns a tuple, the graph and directionality
+	# returns a tuple, (the graph, directionality)
 	nxG, directed = load_graph(graph_file)
 
 	# convert networkx graph to GraphSpace object
@@ -52,15 +56,21 @@ def get_gs_graph(graph_file:str,graph_name:str) -> GSGraph:
 		G.add_node_style(n,color='#ACCE9A',shape='rectangle',width=30,height=30)
 	for u,v in nxG.edges():
 		if directed:
-			G.add_edge(u,v,directed=True,popup='Directed Edge %s-%s<br>Rank %d' % (u,v,nxG[u][v]['rank']))
+			G.add_edge(u,v,directed=True,popup='Directed Edge %s-%s<br>Rank %d' % (u,v,nxG[u][v]['Rank']))
 			G.add_edge_style(u,v,directed=True,width=2,color='#281D6A')
 		else:
-			G.add_edge(u,v,popup='Undirected Edge %s-%s<br>Rank %d' % (u,v,nxG[u][v]['rank']))
+			G.add_edge(u,v,popup='Undirected Edge %s-%s<br>Rank %d' % (u,v,nxG[u][v]['Rank']))
 			G.add_edge_style(u,v,width=2,color='#281D6A')
 	return G
 
 
 def load_graph(path: str) -> nx.Graph:
+	"""
+    Returns a Graph or Digraph, accompanied by a boolean indicating if it's directed.
+
+    This code is compatible only with fully directed or undirected graphs.
+    If neither, an empty Graph will be returned
+    """
 	G = nx.Graph()
 	directed = False
 
@@ -72,14 +82,17 @@ def load_graph(path: str) -> nx.Graph:
 	pathways.columns = ["Interactor1", "Interactor2", "Rank", "Direction"]
 	mask_u = pathways['Direction'] == 'U'
 	mask_d = pathways['Direction'] == 'D'
+	pathways.drop(columns=["Direction"])
 
 	if mask_u.all():
-		G = nx.read_edgelist(path,data=(('rank',float), ('Direction',str)))
+		G = nx.from_pandas_edgelist(pathways, "Interactor1", "Interactor2", ["Rank"])
 		directed = False
+
 	elif mask_d.all():
-		G = nx.read_edgelist(path,data=(('rank',float),('Direction',str)), create_using=nx.DiGraph)
+		G = nx.from_pandas_edgelist(pathways, "Interactor1", "Interactor2", ["rank"], create_using=nx.DiGraph())
 		directed = True
 	else:
-		print("graphspace does not deal with mixed direction type graphs currently")
+		print(f"{path} could not be visualized. GraphSpace does not deal with mixed direction type graphs currently")
+
 
 	return G, directed

spras/analysis/ml.py

@@ -49,7 +49,7 @@ def summarize_networks(file_paths: Iterable[Union[str, PathLike]]) -> pd.DataFra
                     node2 = parts[1]
                     direction = str(parts[3]).strip()
                     if direction == "U":
-                        edges.append(UNDIR_CONST.join([node1, node2]))
+                        edges.append(UNDIR_CONST.join(sorted([node1, node2])))
                     elif direction == "D":
                         edges.append(DIR_CONST.join([node1, node2]))
                     else:
@@ -308,5 +308,10 @@ def ensemble_network(dataframe: pd.DataFrame, output_file: str):
     """
     row_means = dataframe.mean(axis=1, numeric_only=True).reset_index()
     row_means.columns = ['Edges', 'Frequency']
+
+    row_means['Direction'] = row_means['Edges'].apply(lambda edge: 'D' if '-->' in edge else 'U')
+    row_means['Node1'] = row_means['Edges'].apply(lambda edge: edge.split('-->')[0] if '-->' in edge else edge.split('---')[0])
+    row_means['Node2'] = row_means['Edges'].apply(lambda edge: edge.split('-->')[1] if '-->' in edge else edge.split('---')[1])
+
     make_required_dirs(output_file)
-    row_means.to_csv(output_file, sep='\t', index=False, header=True)
+    row_means[['Node1', 'Node2', 'Frequency', "Direction"]].to_csv(output_file, sep='\t', index=False, header=True)

test/ml/expected/expected-dataframe.csv

@@ -8,11 +8,9 @@ O---P,0,0,1,0,0,0,1,0
 P---Q,0,0,1,0,0,0,0,0
 node1---node2,0,0,0,1,0,0,0,0
 node1---node3,0,0,0,1,1,0,0,0
-node4---node5,0,0,0,1,0,0,0,0
-TEST---LONGERNAMES,0,0,0,1,0,0,0,0
-node3---node2,0,0,0,0,1,0,0,0
-node5---node4,0,0,0,0,1,0,0,0
-LONGERNAMES---TEST,0,0,0,0,1,0,0,0
+node4---node5,0,0,0,1,1,0,0,0
+LONGERNAMES---TEST,0,0,0,1,1,0,0,0
+node2---node3,0,0,0,0,1,0,0,0
 nodes with---spaces in name,0,0,0,0,0,0,1,0
 A-->B,0,0,0,0,0,0,0,1
 B-->A,0,0,0,0,0,0,0,1

test/ml/expected/expected-ensemble-network.tsv

@@ -1,10 +1,10 @@
-Edges	Frequency
-A---B	0.5
-C---D	0.75
-E---F	0.75
-L---M	0.5
-M---N	0.25
-O---P	0.25
-P---Q	0.25
-A-->B	0.25
-B-->A	0.25
+Node1	Node2	Frequency	Direction
+A	B	0.5	U
+C	D	0.75	U
+E	F	0.75	U
+L	M	0.5	U
+M	N	0.25	U
+O	P	0.25	U
+P	Q	0.25	U
+A	B	0.25	D
+B	A	0.25	D