Esu sample probability fix

src/esu.py

@@ -8,39 +8,45 @@
 """
 
 import networkx as nx
+import random
 
 
 class ESU:
     def __init__(self, G: nx.Graph):
         self.G = G
 
-    def enumerate_subgraphs(self, size: int):
+    #currently does not allow for motif size over 10 because of default sample probability. should be fixed
+    def enumerate_subgraphs(self, size: int, sample_probability: list[float] = [1,1,1,1,1,1,1,1,1,1]):
         """
         Enumerates all unique subgraphs of a given motif size from the input\
                 graph using the ESU algorithm.
-        Translated for networkx from https://github.com/IlyaVasUW/NEMO_SUITE/
+        Translated and improved for networkx from https://github.com/IlyaVasUW/NEMO_SUITE/
         """
         subgraph_list = []
         nodes = self.G.nodes()
         node_visited = set()
 
-        for node in nodes:
+        sample_neighbors = self.eliminate_percentage(list(nodes), sample_probability[0])
+
+        for node in sample_neighbors:
             neighbor_set = set(self.get_right_neighbors(node))
             node_list = [node]
             node_visited.add(node)
-            self.esu_recursive_helper(size, neighbor_set, node_list, subgraph_list, node_visited)
+            self.esu_recursive_helper(size, neighbor_set, node_list, subgraph_list, node_visited, sample_probability)
 
         return subgraph_list
 
-    def esu_recursive_helper(self, size: int, neighbors: set, node_list: list, subgraph_list: list, nodes_visited: set):
+    def esu_recursive_helper(self, size: int, neighbors: set, node_list: list, subgraph_list: list, nodes_visited: set, sample_probability: list[float]):
         if size == 1:
             subgraph_list.append(self.G.subgraph(node_list.copy()))
             return
 
         if len(neighbors) == 0:
             return
 
-        for node in neighbors:
+        sample_neighbors = self.eliminate_percentage(list(neighbors), sample_probability[len(node_list)])
+
+        for node in sample_neighbors:
             node_list.append(node)
             nodes_visited.add(node)
             next_neighbors = set()
@@ -52,7 +58,7 @@ def esu_recursive_helper(self, size: int, neighbors: set, node_list: list, subgr
                 if neighbor not in nodes_visited:
                     next_neighbors.add(neighbor)
 
-            self.esu_recursive_helper(size - 1, next_neighbors, node_list, subgraph_list, nodes_visited)
+            self.esu_recursive_helper(size - 1, next_neighbors, node_list, subgraph_list, nodes_visited, sample_probability)
 
             if len(node_list) > 0:
                 node_list.pop()
@@ -70,3 +76,16 @@ def get_right_neighbors(self, node):
                 right_hand_neighbors.append(n)
 
         return iter(right_hand_neighbors)
+
+    #Removes a given percentage of elements from a list randomly.
+    #KEEPS PRECENTAGE INPUT. REMOVES 1 - PRECENTAGE
+    def eliminate_percentage(self, collection, probability: float):
+        num_to_remove = int(len(collection) * (1-probability))
+        indices_to_remove = random.sample(range(len(collection)), num_to_remove)
+
+        # Remove elements in reverse order to avoid index issues
+        for i in sorted(indices_to_remove, reverse=True):
+            collection.pop(i)
+
+        return collection
+

src/graph_utils.py

@@ -83,7 +83,7 @@ def draw_graph(self):
     def draw_subgraph(self, motif_size: int):
         output_dir = "drawings/subgraphs"
         esu = ESU(self.G)
-        esu_list = esu.enumerate_subgraphs(motif_size)  # FIXME: 3 is default motif size
+        esu_list = esu.enumerate_subgraphs(motif_size)
 
         for i, subgraph in enumerate(esu_list):
             if self.graph_type == "Directed":
@@ -102,6 +102,6 @@ def draw_subgraph(self, motif_size: int):
             with open(file_name, "r") as f:
                 html = f.read()
 
-            st.markdown(f"### Subgraph {labelGraph(subgraph, self.graph_type)}")
+            st.markdown(f"### Subgraph {i}: {labelGraph(subgraph, self.graph_type)}")
             components.html(html, height=600, scrolling=True)
         return