Merge pull request #15 from solislemuslab/feature/cluster

conserved subgraphs

README.md

@@ -64,6 +64,10 @@ This utility has the form `./minaa.exe <G> <H> [-B=bio] [-a=alpha] [-b=beta]`.
 - **-st=**: similarity threshold; The similarity value above which aligned pairs are included in the output.
   - Require: a real number in range [0, 1].
   - Default: 0.
+- **-c**: conserved subgraphs; whether or not to output a list of the conserved subgraphs in the alignment between G and H.
+  - Require: none.
+  - Default: this list is not calculated or returned.
+  - Note: We define a conserved subgraph as a connected subgraph of G whose nodes are aligned to a connected subgraph of H.
 
 #### Uncommon
 

examples/README.md

@@ -31,12 +31,12 @@ Here we align network **g** with network **h** using topological information and
 ## Example 3
 
 ```bash
-./minaa.exe examples/g.csv examples/h.csv -Galias=nonsmoker -Halias=smoker -p -t
+./minaa.exe examples/g.csv examples/h.csv -Galias=nonsmoker -Halias=smoker -p -t -c
 ```
 
 Output to: `nonsmoker-smoker-2024_01_16-22_05_34/`
 
-Here we align network **g** with network **h**, where **g** is given the alias "nonsmoker", and **h** is given the alias "smoker". The timestamp option `-t` was specified, so the name of the output folder will be nonsmoker-smoker-T, where T is the date and time of execution. Additionally, because the passthrough option `-p` was specified, g.csv and h.csv will be passed through to the output folder as nonsmoker.csv and smoker.csv, respectively.
+Here we align network **g** with network **h**, where **g** is given the alias "nonsmoker", and **h** is given the alias "smoker". The timestamp option `-t` was specified, so the name of the output folder will be nonsmoker-smoker-T, where T is the date and time of execution. Because the passthrough option `-p` was specified, g.csv and h.csv will be passed through to the output folder as nonsmoker.csv and smoker.csv, respectively. Finally, because the `-c` option was specified, the output folder will include the alignment's conserved subgraphs, in a file called `conserved_subgraphs.csv`.
 
 ## Example 4
 

examples/nonsmoker-smoker-2024_01_16-23_10_14/conserved_subgraphs.csv

@@ -0,0 +1,128 @@
+"g1","h146"
+
+"g6","h5"
+
+"g7","h161"
+
+"g8","h138"
+"g167","h134"
+
+"g12","h124"
+
+"g13","h43"
+"g126","h121"
+
+"g14","h171"
+
+"g16","h96"
+
+"g17","h14"
+"g74","h74"
+"g82","h71"
+"g23","h81"
+"g81","h23"
+"g90","h90"
+"g71","h17"
+"g70","h70"
+"g58","h13"
+
+"g19","h25"
+
+"g20","h128"
+"g28","h87"
+
+"g22","h160"
+
+"g24","h162"
+
+"g30","h130"
+
+"g32","h122"
+"g45","h156"
+"g42","h145"
+
+"g35","h135"
+"g155","h120"
+"g164","h60"
+"g112","h109"
+"g92","h68"
+"g51","h104"
+
+"g37","h110"
+
+"g43","h56"
+
+"g46","h155"
+
+"g52","h127"
+
+"g53","h63"
+"g68","h142"
+
+"g54","h123"
+
+"g56","h118"
+
+"g59","h152"
+
+"g63","h129"
+
+"g65","h7"
+
+"g73","h115"
+
+"g91","h55"
+
+"g95","h153"
+
+"g103","h117"
+
+"g105","h132"
+
+"g106","h107"
+
+"g108","h28"
+
+"g109","h93"
+
+"g115","h72"
+
+"g116","h169"
+
+"g119","h103"
+
+"g120","h101"
+
+"g123","h73"
+
+"g125","h157"
+
+"g128","h173"
+"g157","h6"
+
+"g130","h102"
+
+"g131","h51"
+
+"g133","h32"
+
+"g137","h174"
+
+"g139","h30"
+
+"g142","h31"
+
+"g150","h22"
+
+"g154","h21"
+
+"g166","h18"
+
+"g169","h125"
+
+"g172","h44"
+
+"g173","h106"
+
+"g174","h137"
+

include/file_io.h

@@ -21,6 +21,7 @@ namespace FileIO
     void matrix_to_file(std::string, std::vector<std::string>, std::vector<std::string>, std::vector<std::vector<double>>);
     void alignment_to_matrix_file(std::string, std::vector<std::string>, std::vector<std::string>, std::vector<std::vector<double>>, double);
     void alignment_to_list_file(std::string, std::vector<std::string>, std::vector<std::string>, std::vector<std::vector<double>>, double);
+    void subgraphs_to_file(std::string, std::vector<std::string>, std::vector<std::string>, std::vector<std::vector<std::pair<unsigned, unsigned>>>);
 }
 
 #endif

include/util.h

@@ -10,6 +10,7 @@ namespace Util
     std::vector<std::vector<double>> normalize(std::vector<std::vector<double>>);
     std::vector<std::vector<double>> one_minus(std::vector<std::vector<double>>);
     std::vector<std::vector<double>> combine(std::vector<std::vector<double>>, std::vector<std::vector<double>>, double);
+    std::vector<std::vector<std::pair<unsigned, unsigned>>> conserved_subgraphs(std::vector<std::vector<unsigned>>, std::vector<std::vector<unsigned>>, std::vector<std::vector<double>>, double);
 }
 
 #endif

src/file_io.cpp

@@ -584,8 +584,10 @@ namespace FileIO
      *
      * @throws std::runtime_error If the file could not be written.
      */
-    void matrix_to_file(std::string filepath, std::vector<std::string> g_labels, std::vector<std::string> h_labels,
-                        std::vector<std::vector<double>> matrix)
+    void matrix_to_file(std::string filepath, 
+        std::vector<std::string> g_labels, 
+        std::vector<std::string> h_labels,
+        std::vector<std::vector<double>> matrix)
     {
         // Create and open the file
         std::ofstream fout;
@@ -628,8 +630,11 @@ namespace FileIO
      *
      * @throws std::runtime_error If the file could not be written.
      */
-    void alignment_to_matrix_file(std::string filepath, std::vector<std::string> g_labels,
-                                  std::vector<std::string> h_labels, std::vector<std::vector<double>> alignment, double similarity_threshold)
+    void alignment_to_matrix_file(std::string filepath, 
+        std::vector<std::string> g_labels,
+        std::vector<std::string> h_labels,
+        std::vector<std::vector<double>> alignment, 
+        double similarity_threshold)
     {
         // Create and open the file
         std::ofstream fout;
@@ -680,8 +685,11 @@ namespace FileIO
      *
      * @throws std::runtime_error If the file could not be written.
      */
-    void alignment_to_list_file(std::string filepath, std::vector<std::string> g_labels,
-                                std::vector<std::string> h_labels, std::vector<std::vector<double>> alignment, double similarity_threshold)
+    void alignment_to_list_file(std::string filepath, 
+        std::vector<std::string> g_labels,
+        std::vector<std::string> h_labels, 
+        std::vector<std::vector<double>> alignment, 
+        double similarity_threshold)
     {
         // Convert the alignment matrix into a list
         std::vector<std::array<double, 3>> list;
@@ -730,4 +738,53 @@ namespace FileIO
         fout.close();
     }
 
+    /**
+     * Write the given conserved subgraphs to a csv file.
+     *
+     * @param filepath The path to the output file.
+     * @param g_labels Labels for the G graph.
+     * @param h_labels Labels for the H graph.
+     * @param subgraphs The subgraphs to write to the file.
+     *
+     * @throws std::runtime_error If the file could not be written.
+     */
+    void subgraphs_to_file(std::string filepath,
+        std::vector<std::string> g_labels,
+        std::vector<std::string> h_labels, 
+        std::vector<std::vector<std::pair<unsigned, unsigned>>> subgraphs)
+    {
+        // Create and open the file
+        std::ofstream fout;
+        fout.exceptions(std::ofstream::badbit);
+        try
+        {
+            fout.open(filepath);
+        }
+        catch (const std::ofstream::failure &e)
+        {
+            throw std::runtime_error("Unable to open file " + filepath);
+        }
+
+        // Write the subgraphs to the file
+        for (const auto& subgraph : subgraphs)
+        {
+            for (const auto& pair : subgraph)
+            {
+                unsigned g_index = pair.first;
+                unsigned h_index = pair.second;
+
+                // Ensure the indices are within the bounds of the labels
+                if (g_index >= g_labels.size() || h_index >= h_labels.size())
+                {
+                    throw std::runtime_error("Index out of bounds for labels.");
+                }
+
+                fout << g_labels[g_index] << "," << h_labels[h_index] << std::endl;
+            }
+            fout << std::endl;
+        }
+
+        fout.close();
+    }
+
 }

src/minaa.cpp

@@ -45,10 +45,11 @@ int main(int argc, char *argv[])
         auto g_alias = args[7];                            // graph G alias
         auto h_alias = args[8];                            // graph H alias
         auto bio_alias = args[9];                          // biological data alias
-        auto do_passthrough = (args[10] == "1");           // do a passthrough of input files?
-        auto do_timestamp = (args[11] == "1");             // include a timestamp in the directory name?
-        auto do_greekstamp = (args[12] == "1");            // include a greekstamp in the directory name?
-        auto do_similarity_conversion = (args[13] == "1"); // convert biological similarity to costs?
+        auto do_subgraph_id = (args[10] == "1");           // do conserved subgraph identification?
+        auto do_passthrough = (args[11] == "1");           // do a passthrough of input files?
+        auto do_timestamp = (args[12] == "1");             // include a timestamp in the directory name?
+        auto do_greekstamp = (args[13] == "1");            // include a greekstamp in the directory name?
+        auto do_similarity_conversion = (args[14] == "1"); // convert biological similarity to costs?
         auto do_bio = (bio_file != "");                    // biological data file provided?
 
         const auto BASE_PATH = "alignments";
@@ -58,6 +59,7 @@ int main(int argc, char *argv[])
         const auto OVERALL_COSTS_FILENAME = "overall_costs.csv";
         const auto ALIGNMENT_MATRIX_FILENAME = "alignment_matrix.csv";
         const auto ALIGNMENT_LIST_FILENAME = "alignment_list.csv";
+        const auto CONSERVED_SUBGRAPHS_FILENAME = "conserved_subgraphs.csv";
 
         // Generate output names
         auto g_name = FileIO::name_file(g_file, g_alias);
@@ -235,6 +237,17 @@ int main(int argc, char *argv[])
         auto d51 = std::chrono::duration_cast<std::chrono::milliseconds>(f51 - s51).count();
         FileIO::out(log, "done. (" + std::to_string(d51) + "ms)\n");
 
+        if (do_subgraph_id) {
+            // Conserved subgraph identification
+            FileIO::out(log, "Identifying conserved subgraphs................");
+            auto s60 = std::chrono::high_resolution_clock::now();
+            auto subgraphs = Util::conserved_subgraphs(g_graph, h_graph, alignment, similarity_threshold);
+            FileIO::subgraphs_to_file(directory + CONSERVED_SUBGRAPHS_FILENAME, g_labels, h_labels, subgraphs);
+            auto f60 = std::chrono::high_resolution_clock::now();
+            auto d60 = std::chrono::duration_cast<std::chrono::milliseconds>(f60 - s60).count();
+            FileIO::out(log, "done. (" + std::to_string(d60) + "ms)\n");
+        }
+
         auto f = std::chrono::high_resolution_clock::now();
         auto d = std::chrono::duration_cast<std::chrono::milliseconds>(f - s).count();
         FileIO::out(log, "ALIGNMENT COMPLETED (" + std::to_string(d) + "ms)\n");