Big cleanup for JSON format support

Louis BECQUEY
Commit 89aa05565af08c7b5a42acfd2542a0ce6183bd23 89aa0556 1 parent 3b2bbb99
Showing 11 changed files with 47 additions and 107 deletions
Isaure_benchmark.py
Makefile
cppsrc/MOIP.cpp
cppsrc/MOIP.h
cppsrc/Motif.cpp
cppsrc/Motif.h
cppsrc/SecondaryStructure.cpp
cppsrc/SecondaryStructure.h
cppsrc/biorseo.cpp
cppsrc/program
cppsrc/rna.cpp
--- a/Isaure_benchmark.py deleted 100644 → 0
View file @3b2bbb9
+++ b/Isaure_benchmark.py deleted 100644 → 0
View file @3b2bbb9
--- a/Makefile
View file @89aa055
+++ b/Makefile
View file @89aa055
@@ -40,12 +40,9 @@ re: remove clean all
 .PHONY: clean
 clean:
 	$(rm) $(OBJECTS)
-	$(rm) doc/supplementary_material.bbl doc/supplementary_material.blg doc/supplementary_material.synctex.gz doc/supplementary_material.log doc/supplementary_material.aux
-	$(rm) doc/main_bioinformatics.bbl doc/main_bioinformatics.blg doc/main_bioinformatics.synctex.gz doc/main_bioinformatics.log doc/main_bioinformatics.aux doc/OUP_First_SBk_Bot_8401-eps-converted-to.pdf
 	@echo -e "\033[00;32mCleanup completed.\033[00m"
 .PHONY: remove
 remove:
 	@$(rm) $(BINDIR)/$(TARGET)
-	@$(rm) doc/main_bioinformatics.pdf doc/supplementary_material.pdf
 	@echo -e "\033[00;32mExecutable and docs removed!\033[00m"
--- a/cppsrc/MOIP.cpp
View file @89aa055
+++ b/cppsrc/MOIP.cpp
View file @89aa055
--- a/cppsrc/MOIP.h
View file @89aa055
+++ b/cppsrc/MOIP.h
View file @89aa055
@@ -12,12 +12,19 @@
 using std::vector;
-typedef struct args_ {
+typedef struct argsf_ {
 						path           motif_file;
 						std::mutex&    posInsertionSites_mutex;
-						args_(path motif_file_, mutex& mutex_) : motif_file(motif_file_), posInsertionSites_mutex(mutex_) {}
+						argsf_(path motif_file_, mutex& mutex_) 
-					  } args_of_parallel_func;
+							: motif_file(motif_file_), posInsertionSites_mutex(mutex_) {}
+					  } file_and_mutex;
+typedef struct argsm_ {
+						json_elem      motif;
+						std::mutex&    posInsertionSites_mutex;
+						argsm_(json_elem& motif_, mutex& mutex_) 
+							: motif(motif_), posInsertionSites_mutex(mutex_) {}
+					  } motif_and_mutex;
 class MOIP
 {
@@ -56,9 +63,9 @@ class MOIP
 	bool   						exists_vertical_outdated_labels(const SecondaryStructure& s) const;
 	bool   						exists_horizontal_outdated_labels(const SecondaryStructure& s) const;
-	void   						allowed_motifs_from_desc(args_of_parallel_func arg_struct);
+	void   						allowed_motifs_from_desc(file_and_mutex arg_struct);
-	void   						allowed_motifs_from_rin(args_of_parallel_func arg_struct);
+	void   						allowed_motifs_from_rin(file_and_mutex arg_struct);
-	void						allowed_motifs_from_json(args_of_parallel_func arg_struct, vector<pair<uint, char>> errors_id);
+	void						allowed_motifs_from_json(motif_and_mutex arg_struct);
 	bool verbose_;    // Should we print things ?
@@ -79,7 +86,6 @@ class MOIP
 	vector<vector<size_t>> index_of_Cxip_;               // Stores the indexes of the Cxip in insertion_dv_
 	vector<size_t>         index_of_first_components;    // Stores the indexes of Cx1p in insertion_dv_
 	vector<vector<size_t>> index_of_yuv_;                // Stores the indexes of the y^u_v in basepair_dv_
-
 	vector<vector<size_t>> index_of_xij_;		         //Stores the indexes of the xij variables (BioKop) in stacks_dv_
 };
--- a/cppsrc/Motif.cpp
View file @89aa055
+++ b/cppsrc/Motif.cpp
View file @89aa055
--- a/cppsrc/Motif.h
View file @89aa055
+++ b/cppsrc/Motif.h
View file @89aa055
@@ -7,6 +7,7 @@
 #include <vector>
 #include <filesystem>
 #include "rna.h"
+#include "json.hpp"
 using boost::filesystem::path;
 using std::pair;
@@ -14,6 +15,8 @@ using std::string;
 using std::vector;
 using std::mutex;
+typedef enum { RNA3DMOTIF = 1, CSV = 2, CARNAVAL = 3, JSON = 4 } source_type;
+typedef nlohmann::detail::iter_impl<nlohmann::basic_json<> > json_elem;
 typedef struct Comp_ {
@@ -42,15 +45,14 @@ class Motif
   public:
     Motif(void);
     Motif(string csv_line);
-    Motif(const vector<Component>& v, string PDB);
+    Motif(const vector<Component>& v, string name);
-    Motif(const vector<Component>& v, string id, size_t contacts, double tx_occurrences);
+    Motif(const vector<Component>& v, string name, string& struc);
     Motif(const vector<Component>& v, path rinfile, uint id, bool reversed);
+    // Motif(string path, int id); //full path to biorseo/data/modules/RIN/Subfiles/
-
-    Motif(string path, int id); //full path to biorseo/data/modules/RIN/Subfiles/
     static char       is_valid_RIN(const string& rinfile);
     static char       is_valid_DESC(const string& descfile);
-    static vector<pair<uint,char>>          is_valid_JSON(const string& jsonfile);
+    static char       is_valid_JSON(const json_elem& i);
     string            pos_string(void) const;
     string            sec_struct(void) const;
@@ -64,39 +66,27 @@ class Motif
     double            tx_occurrences_;
     double            score_;
     bool              reversed_;
+    static uint       delay;
+    // delay is the minimal shift between end of a component and begining of the next.
+    // For regular loop motifs, it should be at least 5 (because hairpins cannot be of size smaller than 5).
+    // For the general case, it could be zero, but solutions will look dirty...
+    // Higher values reduce combinatorial explosion of potential insertion sites.
   private:
-    string carnaval_id;  // if source = CARNAVAL
+    string id_;
-    string atlas_id;     // if source = RNAMOTIFATLAS
+    source_type source_;
-    string PDBID;        // if source = RNA3DMOTIF
-    string contacts_id;  // if source = CONTACTS
-    bool   is_model_;    // Whether the motif is a model or an extracted module from a 3D structure
-    enum { RNA3DMOTIF = 1, RNAMOTIFATLAS = 2, CARNAVAL = 3, CONTACTS = 4 } source_;
 };
 bool                        is_desc_insertible(const string& descfile, const string& rna);
-bool                        is_rin_insertible(const string& rinfile, const string& rna);
+bool                        check_motif_ss(string);
-bool                        is_json_insertible(const string& jsonfile, const string& rna);
+bool                        check_motif_sequence(string);
 vector<Motif>               load_txt_folder(const string& path, const string& rna, bool verbose);
 vector<Motif>               load_desc_folder(const string& path, const string& rna, bool verbose);
 vector<Motif>               load_csv(const string& path);
 vector<Motif>               load_json_folder(const string& path, const string& rna, bool verbose);
-vector<vector<Component>>   find_next_ones_in(string rna, uint offset, vector<string>& vc);
+vector<vector<Component>>   find_next_ones_in(string rna, uint offset, vector<string> vc);
-vector<vector<Component>>   json_find_next_ones_in(string rna, uint offset, vector<string>& vc);
-
-// utilities for Json motifs
-size_t count_nucleotide(string&);
-size_t count_delimiter(string&);
-size_t count_contacts(string&);
-string check_motif_sequence(string);
-bool checkSecondaryStructure(string);
-vector<Link> build_motif_pairs(string&, vector<Component>&);
-uint find_max_occurrences(string&);
-uint find_max_sequence(string&);
-vector<string> find_components(string&, string);
-vector<uint> find_contacts(vector<string>&, vector<Component>&);
 // utilities to compare secondary structures:
 bool operator==(const Motif& m1, const Motif& m2);
--- a/cppsrc/SecondaryStructure.cpp
View file @89aa055
+++ b/cppsrc/SecondaryStructure.cpp
View file @89aa055
@@ -3,16 +3,12 @@
 #include <algorithm>
 #include <boost/format.hpp>
-#define RESET   "\033[0m"
-#define RED     "\033[31m"      /* Red */
-
 using std::abs;
 using std::cout;
 using std::endl;
 SecondaryStructure::SecondaryStructure() {}
-
 SecondaryStructure::SecondaryStructure(const RNA& rna)
 : objective_scores_(vector<double>(2)), n_(rna.get_RNA_length()), nBP_(0), rna_(rna)
 {
@@ -21,8 +17,6 @@ SecondaryStructure::SecondaryStructure(const RNA& rna)
 SecondaryStructure::SecondaryStructure(bool empty) : rna_(RNA()) { is_empty_structure = empty; }
-
-
 string SecondaryStructure::to_DBN(void) const
 {
@@ -100,26 +94,6 @@ string SecondaryStructure::to_DBN(void) const
     return res;
 }
-string structure_with_contacts(const SecondaryStructure& ss) {
-    string sequence = ss.rna_.get_seq();
-    string construct = "";
-    bool flag;
-    for (uint i = 0; i < sequence.size(); i++) {
-        flag = false;
-        for (const Motif& m : ss.motif_info_) {
-            for (uint j = 0; j < m.pos_contacts.size(); j++) {
-                if (m.pos_contacts[j] == i) flag = true;
-            }
-        }
-        if (flag) {   
-            construct += "*";
-        } else {
-            construct += ".";
-        }
-    }
-    return construct;
-}
-
 string SecondaryStructure::to_string(void) const
 {
     string s;
@@ -141,35 +115,11 @@ void SecondaryStructure::set_basepair(uint i, uint j)
 void SecondaryStructure::insert_motif(const Motif& m) { motif_info_.push_back(m); }
-void colored_contacts(string sequence, vector<Motif> motif_info_) {
-    bool flag;
-    for (uint i = 0; i < sequence.size(); i++) {
-        flag = false;
-        for (const Motif& m : motif_info_) {
-            for (uint j = 0; j < m.pos_contacts.size(); j++) {
-                if (m.pos_contacts[j] == i) flag = true;
-            }
-        }
-        if (flag) {   
-            cout << RED << sequence[i] << RESET;
-        } else {
-            cout << sequence[i];
-        }
-    }
-}
-
 void SecondaryStructure::print(void) const
 {
-    cout << endl;
+    cout << endl << '\t' << rna_.get_seq() << endl;
-    cout << '\t';
-    colored_contacts(rna_.get_seq(), motif_info_);
-    //rna_.get_seq() 
-    cout << endl;
     string ss = to_string();
-    cout << '\t';
+    cout << '\t' << ss << endl;
-    colored_contacts(ss, motif_info_);
-    //cout << ss;
-    cout << endl;
     for (const Motif& m : motif_info_) {
         uint i = 0;
         cout << '\t';
@@ -324,7 +274,6 @@ bool operator<=(const SecondaryStructure& s1, const SecondaryStructure& s2)
     return false;
 }
-
 bool operator==(const SecondaryStructure& s1, const SecondaryStructure& s2)
 {
     // Checks wether the secondary structures are exactly the same, including the inserted motifs.
--- a/cppsrc/SecondaryStructure.h
View file @89aa055
+++ b/cppsrc/SecondaryStructure.h
View file @89aa055
@@ -57,7 +57,4 @@ inline void   SecondaryStructure::set_objective_score(int i, double s) { objecti
 inline uint   SecondaryStructure::get_n_motifs(void) const { return motif_info_.size(); }
 inline uint   SecondaryStructure::get_n_bp(void) const { return nBP_; }
-string structure_with_contacts(const SecondaryStructure& ss);
-
-
 #endif    //  SECONDARY_STRUCTURE_
\ No newline at end of file
--- a/cppsrc/biorseo.cpp
View file @89aa055
+++ b/cppsrc/biorseo.cpp
View file @89aa055
@@ -79,12 +79,10 @@ int main(int argc, char* argv[])
     ("jsonfolder,j", po::value<string>(&motifs_path_name), "A folder containing a custom motif library in .json format")
     ("pre-placed,x", po::value<string>(&motifs_path_name), "A CSV file providing motif insertion sites obtained with another tool.")
     ("function,f", po::value<char>(&obj_function_nbr)->default_value('B'), 
-                    "(A, B, C, D, E or F) Objective function to score module insertions:\n"
+                    "(A, B, C, or D) Objective function to score module insertions:\n"
                     "  (A) insert big modules\n  (B) light, high-order modules\n"
                     "  (C) well-scored modules\n  (D) light, high-order, well-scored\n    modules\n"
-                    "  (E, F) insert big modules with many\n   contacts with proteins, different\n   ponderations.\n"
+                    "  C and D require position scores\n  provided by --pre-placed.\n")    
-                    "  C and D require position scores\n  provided by --pre-placed.\n"
-                    "  E and F require protein-contact\n  information and should be\n  used only with --jsonfolder.")    
     ("mfe,E", "Minimize stacking energies\n  (leads to MFE extimator)")
     ("mea,A", "(default) Maximize expected accuracy\n  (leads to MEA estimator)")
     ("first-objective,c",   po::value<unsigned int>(&MOIP::obj_to_solve_)->default_value(2), 
@@ -108,8 +106,8 @@ int main(int argc, char* argv[])
                 << "Bio-objective integer linear programming framework to predict RNA secondary structures by including known RNA modules." << endl
                 << "Developped by Louis Becquey, 2018-2021\nLénaïc Durand, 2019\nNathalie Bernard, 2021" << endl << endl
                 << "Usage:\tYou must provide:\n\t1) a FASTA input file with -s," << endl
-                << "\t2) a module type with --rna3dmotifs, --carnaval, --contacts or --pre-placed," << endl
+                << "\t2) a module type with --rna3dmotifs, --carnaval, --json or --pre-placed," << endl
-                << "\t3) one module-based scoring function with --func A, B, C, D, E or F," << endl
+                << "\t3) one module-based scoring function with --func A, B, C, or D" << endl
                 << "\t4) one energy-based scoring function with --mfe or --mea," << endl
                 << "\t5) how to display results: in console (-v), or in a result file (-o)." << endl
                 << endl
@@ -154,13 +152,14 @@ int main(int argc, char* argv[])
         return EXIT_FAILURE;
     }
-
-    /*  FIND PARETO SET  */
     string source;
+    Motif::delay = 1;
     if (vm.count("rinfolder"))
         source = "rinfolder";
-    else if (vm.count("descfolder"))
+    else if (vm.count("descfolder")) {
         source = "descfolder";
+        Motif::delay = 5;
+    }
     else if (vm.count("jsonfolder"))
         source = "jsonfolder";
     else if (vm.count("pre-placed"))
@@ -168,6 +167,8 @@ int main(int argc, char* argv[])
     else
         cerr << "ERR: no source of modules provided !" << endl;
+    /*  FIND PARETO SET  */
+    
     MOIP               myMOIP = MOIP(myRNA, source, motifs_path_name.c_str(), theta_p_threshold, verbose);
     double             min, max;
     IloConstraintArray F(myMOIP.get_env());
@@ -243,11 +244,8 @@ int main(int argc, char* argv[])
         outfile.open(outputName);
         outfile << fa->name() << endl << fa->seq() << endl;
-        for (uint i = 0; i < myMOIP.get_n_solutions(); i++) {
+        for (uint i = 0; i < myMOIP.get_n_solutions(); i++)
-            outfile << myMOIP.solution(i).to_string() << endl << structure_with_contacts(myMOIP.solution(i)) << endl;
+            outfile << myMOIP.solution(i).to_string() << endl;
-            string str1 = myMOIP.solution(i).to_string();
-
-        }
         outfile.close();
     }
--- a/cppsrc/program deleted 100644 → 0
View file @3b2bbb9
+++ b/cppsrc/program deleted 100644 → 0
View file @3b2bbb9
--- a/cppsrc/rna.cpp
View file @89aa055
+++ b/cppsrc/rna.cpp
View file @89aa055
@@ -50,6 +50,7 @@ RNA::RNA(string name, string seq, bool verbose)
 	int max_bp_span = 100;
 	float cutoff = 1e-6;
 	vrna_ep_t* results = vrna_pfl_fold(cseq, window_size, max_bp_span, cutoff);
+	vrna_ep_t* save = results; // keep the pointer to free it later
 	if (results != NULL)
 	{
@@ -96,6 +97,8 @@ RNA::RNA(string name, string seq, bool verbose)
 			}
 		}
+		// Free memory allocated by ViennaRNA
+		free(save);
 	}
 	else cerr << "NULL result returned by vrna_pfl_fold" << endl;