Merge branch '0.1.0' into master

phylo42 · Jun 5, 2019 · be4bdd5 · be4bdd5
2 parents 3085f73 + e1f97e2
commit be4bdd5
Show file tree

Hide file tree

Showing 18 changed files with 612 additions and 403 deletions.
diff --git a/CHANGELOG.txt b/CHANGELOG.txt
@@ -0,0 +1,2 @@
+0.1.0
+First released version.
diff --git a/build/CMakeLists.txt b/build/CMakeLists.txt
@@ -23,9 +23,9 @@ target_sources(build_n
 
             # posterior probability matrix
             src/pp_matrix/proba_matrix.cpp
+            src/pp_matrix/node_entry.cpp
+            src/pp_matrix/node_entry_view.cpp
             src/pp_matrix/phyml.cpp
-            src/pp_matrix/branch_entry.cpp
-            src/pp_matrix/branch_entry_view.cpp
 
             src/main.cpp
         )

diff --git a/build/src/build/db_builder.cpp b/build/src/build/db_builder.cpp
diff --git a/build/src/build/db_builder.h b/build/src/build/db_builder.h
@@ -11,8 +11,9 @@ namespace core
 namespace rappas
 {
     core::phylo_kmer_db build(const std::string& working_directory, const std::string& ar_probabilities_file,
-        const std::string& tree_file, const std::string& extended_mapping_file,
-        const std::string& artree_mapping_file, size_t kmer_size);
+        const std::string& original_tree_file, const std::string& extended_tree_file,
+        const std::string& extended_mapping_file, const std::string& artree_mapping_file,
+        size_t kmer_size, core::phylo_kmer::score_type omega, size_t num_threads);
 }
 
 #endif
diff --git a/build/src/cli/command_line.cpp b/build/src/cli/command_line.cpp
@@ -27,9 +27,12 @@ namespace cli
     static std::string WORKING_DIR = "workdir", WORKING_DIR_SHORT = "w";
     static std::string AR_PROBABILITIES = "ar-probabilities", AR_PROBABILITIES_SHORT = "a";
     static std::string REFTREE = "reftree", REFTREE_SHORT = "t";
+    static std::string EXTENDED_TREE = "extended_tree", EXTENDED_TREE_SHORT = "x";
     static std::string EXTENDED_MAPPING = "extended_mapping", EXTENDED_MAPPING_SHORT = "e";
     static std::string ARTREE_MAPPING = "artree_mapping", ARTREE_MAPPING_SHORT = "m";
     static std::string K = "k", K_SHORT = "k";
+    static std::string OMEGA="omega", OMEGA_SHORT="o";
+    static std::string NUM_THREADS = "num_threads", NUM_THREADS_SHORT = "j";
 
     const po::options_description get_opt_description()
     {
@@ -42,13 +45,19 @@ namespace cli
             ((AR_PROBABILITIES + "," + AR_PROBABILITIES_SHORT).c_str(), po::value<fs::path>()->required(),
              "Ancestral reconstruction probabilities file")
             ((REFTREE + "," + REFTREE_SHORT).c_str(), po::value<fs::path>()->required(),
-             "Phylogenetic tree file")
+             "Original phylogenetic tree file")
+            ((EXTENDED_TREE + "," + EXTENDED_TREE_SHORT).c_str(), po::value<fs::path>()->required(),
+             "Extended phylogenetic tree file")
             ((EXTENDED_MAPPING + "," + EXTENDED_MAPPING_SHORT).c_str(), po::value<fs::path>()->required(),
              "Original mapping file")
             ((ARTREE_MAPPING + "," + ARTREE_MAPPING_SHORT).c_str(), po::value<fs::path>()->required(),
              "Ancestral reconstruction tree mapping file")
             ((K + "," + K_SHORT).c_str(), po::value<size_t>()->default_value(8),
-             "k-mer length used at DB build");
+             "k-mer length used at DB build")
+            ((OMEGA + "," + OMEGA_SHORT).c_str(), po::value<core::phylo_kmer::score_type>()->default_value(1.0),
+             "Score threshold parameter")
+            ((NUM_THREADS + "," + NUM_THREADS_SHORT).c_str(), po::value<size_t>()->default_value(1),
+             "Number of threads");
         return desc;
     }
 
@@ -81,10 +90,13 @@ namespace cli
 
             parameters.working_directory = vm[WORKING_DIR].as<fs::path>().string();
             parameters.ar_probabilities_file = vm[AR_PROBABILITIES].as<fs::path>().string();
-            parameters.tree_file = vm[REFTREE].as<fs::path>().string();
+            parameters.original_tree_file = vm[REFTREE].as<fs::path>().string();
+            parameters.extended_tree_file = vm[EXTENDED_TREE].as<fs::path>().string();
             parameters.extended_mapping_file = vm[EXTENDED_MAPPING].as<fs::path>().string();
             parameters.artree_mapping_file = vm[ARTREE_MAPPING].as<fs::path>().string();
             parameters.kmer_size = vm[K].as<size_t>();
+            parameters.omega = vm[OMEGA].as<core::phylo_kmer::score_type>();
+            parameters.num_threads = vm[NUM_THREADS].as<size_t>();
         }
         catch (const po::error& e)
         {

diff --git a/build/src/cli/command_line.h b/build/src/cli/command_line.h
@@ -4,6 +4,7 @@
 #include <exception>
 #include <string>
 #include <map>
+#include <core/phylo_kmer.h>
 
 namespace cli
 {
@@ -19,12 +20,15 @@ namespace cli
         action_t action;
         std::string working_directory;
         std::string ar_probabilities_file;
-        std::string tree_file;
+        std::string original_tree_file;
+        std::string extended_tree_file;
         std::string extended_mapping_file;
         std::string artree_mapping_file;
 
         // algo options
         size_t kmer_size;
+        core::phylo_kmer::score_type omega;
+        size_t num_threads;
     };
 
     const std::string get_option_list();

diff --git a/build/src/main.cpp b/build/src/main.cpp
@@ -1,12 +1,15 @@
 #include <iostream>
+#include <sstream>
+#include <chrono>
 #include <boost/filesystem.hpp>
 #include <chrono>
 #include <core/phylo_kmer_db.h>
 #include <core/serialization.h>
+#include <iomanip>
 #include "cli/command_line.h"
 #include "cli/exceptions.h"
-#include "return.h"
 #include "build/db_builder.h"
+#include "return.h"
 
 namespace fs = boost::filesystem;
 
@@ -19,6 +22,18 @@ return_code print_help()
     return return_code::help;
 }
 
+std::string generate_db_name(const core::phylo_kmer_db& db)
+{
+    const auto kmer_size = db.kmer_size();
+    const auto omega = db.omega();
+    const auto threshold = core::score_threshold(omega, kmer_size);
+
+    std::ostringstream out;
+    out << "DB_k" << kmer_size << "_w" << std::fixed << std::setprecision(2) << omega << "_t" <<
+        std::setprecision(4) << threshold << ".rps";
+    return out.str();
+}
+
 return_code run(const cli::cli_parameters& parameters)
 {
     switch (parameters.action)
@@ -29,11 +44,18 @@ return_code run(const cli::cli_parameters& parameters)
         }
         case cli::build:
         {
+            if (parameters.kmer_size > core::seq_traits::max_kmer_length)
+            {
+                std::cerr << "Maximum k-mer size allowed: " << core::seq_traits::max_kmer_length << std::endl;
+                return return_code::argument_error;
+            }
+
             const auto db = rappas::build(parameters.working_directory, parameters.ar_probabilities_file,
-                                          parameters.tree_file, parameters.extended_mapping_file,
-                                          parameters.artree_mapping_file, parameters.kmer_size);
+                                          parameters.original_tree_file, parameters.extended_tree_file,
+                                          parameters.extended_mapping_file, parameters.artree_mapping_file,
+                                          parameters.kmer_size, parameters.omega, parameters.num_threads);
 
-            const auto db_filename = fs::path(parameters.working_directory) / "DB.union";
+            const auto db_filename = fs::path(parameters.working_directory) / generate_db_name(db);
 
             std::cout << "Saving database to: " << db_filename.string() << "..." << std::endl;
             std::chrono::steady_clock::time_point begin = std::chrono::steady_clock::now();
@@ -47,7 +69,7 @@ return_code run(const cli::cli_parameters& parameters)
         {
             return return_code::unknown_error;
         }
-    };
+    }
 }
 
 int main(int argc, const char* argv[])
@@ -61,23 +83,24 @@ int main(int argc, const char* argv[])
     }
     catch (const conflicting_options& e)
     {
-        std::cout << e.what() << std::endl;
+        std::cerr << e.what() << std::endl;
         return 1;
     }
     catch (const bad_options& e)
     {
-        std::cout << e.what() << std::endl;
+        std::cerr << e.what() << std::endl;
         return 1;
     }
     catch (const std::runtime_error& e)
     {
-        std::cout << e.what() << std::endl;
+        std::cerr << e.what() << std::endl;
         return 1;
     }
+    /*
     catch (...)
     {
-        std::cout << "Unexpected error. " << std::endl;
+        std::cerr << "Unexpected error. " << std::endl;
         return 1;
-    }
+    }*/
     return 0;
 }
diff --git a/build/src/pp_matrix/branch_entry_view.cpp b/build/src/pp_matrix/branch_entry_view.cpp