Safe script for merging a molecule graph into a uniform barcode graph

deconvolution/generate_fake_barcode_graph.py

 #!/usr/bin/env python3
 
 import networkx  as nx
-import sys
+import random
+import argparse
 
+import graph_manipulator as gm
 
-G = nx.read_graphml (sys.argv[1])
 
-# label molecule nodes
-labels = {}
-for idx, node in enumerate(G.nodes()):
-    labels[node] = str(idx)
+def parse_arguments():
+    parser = argparse.ArgumentParser(description='Transform a 10X molecule graph into a 10X barcode graph.')
+    parser.add_argument('--merging_depth', '-m', type=int, required=True, help='Number of nodes to merge together')
+    parser.add_argument('--input_graph', '-i', required=True, help='A 10X molecule graph gexf formated.')
+    parser.add_argument('--output', '-o', help="Output filename")
+    parser.add_argument('--random_seed', '-s', type=int, help="If you want to fix the random seed for reproducibility")
 
-# artificially make barcodes
-barcodes = []
-n = len(G.nodes())
-available_molecules = set(G.nodes())
-m = 3 # m molecules per barcode
+    args = parser.parse_args()
+    return args
 
-# Group molecules by barcode
-import random
-while len(available_molecules) > 0:
-    barcode = set(random.sample(available_molecules, m))
-    available_molecules -= barcode
-    # print(barcode)
-    barcodes += [barcode]
-
-# Associate molecule to barcode
-molecule_barcode = dict()
-for barcode_index, barcode in enumerate(barcodes):
-    for mol in barcode:
-        molecule_barcode[mol] = barcode_index
-
-print(molecule_barcode)
-
-# Generate barcoded graph nodes
-G2 = nx.Graph()
-g2_labels = {}
-for barcode_index, barcode_molecules in enumerate(barcodes):
-    bar_names = "_".join(barcode_molecules)
-    g2_labels[barcode_index] = f"{barcode_index}:{bar_names}"
-    G2.add_node(g2_labels[barcode_index])
-
-# Generate barcoded graph edges
-for mol_edge in G.edges():
-    m1, m2 = mol_edge
-    b1, b2 = g2_labels[molecule_barcode[m1]], g2_labels[molecule_barcode[m2]]
-    G2.add_edge(b1,b2)
-
-# print(G2.edges)
-
-output = sys.argv[1].replace("molecule", "barcode").replace(".graphml", f"_{m}.gexf")
-nx.write_gexf(G2, output)
 
+""" Take a molecule d-graph chain and merge the nodes uniformly to obtain a barcode graph.
+    @param G A molecule graph
+    @param merging_depth The number of nodes to merge from the original graph to obtain one node of the barcode graph
+    @return The merged barcode graph
+"""
+def fusion_graph(G, merging_depth):
+    nodes = list(G.nodes())
+    random.shuffle(nodes)
+
+    label = 0
+    bijective_labels = {}
+
+    for idx in range(0, len(nodes), merging_depth):
+        # Extract values to merge
+        sublist = nodes[idx : idx+merging_depth]
+        # Merge nodes
+        merged = sublist[0]
+        for sub_idx in range(1, len(sublist)):
+            merged = gm.merge_nodes(G, merged, sublist[sub_idx])
+
+        # Label the node
+        bijective_labels[merged] = f"{label}:{merged}"
+        label += 1
+
+    # Relabel all the nodes
+    G = nx.relabel_nodes(G, bijective_labels)
+
+    return G
+
+
+def save_graph(G, outfile):
+    nx.write_gexf(G, outfile)
+
+
+if __name__ == "__main__":
+    args = parse_arguments()
+
+    if args.random_seed:
+        random.seed(args.random_seed)
+
+    G = nx.read_gexf(args.input_graph)
+    G = fusion_graph(G, args.merging_depth)
+
+    outfile = f"simulated_barcodes_{args.merging_depth}.gexf"
+    if args.output:
+        outfile = args.output
+    save_graph(G, outfile)

deconvolution/generate_fake_molecule_graph.py

@@ -31,6 +31,6 @@ if __name__ == "__main__":
 
     outfile = f"simulated_molecules_{args.num_molecule}_{args.depth}.gexf"
     if args.output:
-        outfile = args.outfile
+        outfile = args.output
     save_graph(G, outfile)
 

deconvolution/graph_manipulator.py

@@ -27,7 +27,7 @@ def generate_d_graph_chain(size, d):
     @param G The graph to manipulate
     @param node1 First node to merge
     @param node2 Second node to merge
-    @return The modified graph G
+    @return The name of the new node in G
 """
 def merge_nodes(G, node1, node2):
     # Create the new node
@@ -63,5 +63,5 @@ def merge_nodes(G, node1, node2):
     G.remove_node(node1)
     G.remove_node(node2)
 
-    return G
+    return new_node