Ajout de fonctions pour faire une liste des motifs qui ne match qu'avec leurs séquence d'origine

@@ -17,7 +17,7 @@ That script will remove from the library all the pattern that match ONLY with th
 
 
 vector<string> get_list_pdb_benchmark(const string& benchmark) {
 vector<string> get_list_pdb_benchmark(const string& benchmark) {
 
 
-    ifstream bm(benchmark);
+    fstream bm(benchmark);
     vector<string> list_pdb;
     vector<string> list_pdb;
     if (bm.is_open()) {
     if (bm.is_open()) {
         string name;
         string name;
@@ -27,7 +27,7 @@ vector<string> get_list_pdb_benchmark(const string& benchmark) {
 
 
         while (getline(bm, name)) {
         while (getline(bm, name)) {
             int size = name.size();
             int size = name.size();
-            name = name.substr(5,size-8); 
+            name = name.substr(5,size-6); 
             list_pdb.push_back(name);
             list_pdb.push_back(name);
 
 
             getline(bm, sequence);
             getline(bm, sequence);
@@ -45,18 +45,20 @@ string trim(string str) {
     return str;
     return str;
 }
 }
 
 
-bool find_id_pattern(string& pdb_pattern, const string& benchmark) {
+string find_id_pattern(string& pdb_pattern, const string& benchmark) {
     vector<string> l = get_list_pdb_benchmark(benchmark);
     vector<string> l = get_list_pdb_benchmark(benchmark);
     for (string pdb_bm : l) {
     for (string pdb_bm : l) {
-        if (!pdb_bm.compare(pdb_pattern)) {
-            return true;
+        int size = pdb_bm.size();
+        string cmp = pdb_bm.substr(0, size-2);
+        if (!cmp.compare(pdb_pattern)) {
+            return pdb_bm;
         }
         }
     }
     }
-    return false;
+    return string();
 }
 }
 
 
 vector<pair<string, string>> find_id(const string& bibli, const string& benchmark) {
 vector<pair<string, string>> find_id(const string& bibli, const string& benchmark) {
-    std::ifstream lib(bibli);
+    ifstream lib(bibli);
     json js = json::parse(lib);
     json js = json::parse(lib);
 
 
     vector<pair<string, string>> association;
     vector<pair<string, string>> association;
@@ -71,9 +73,10 @@ vector<pair<string, string>> find_id(const string& bibli, const string& benchmar
                     ostringstream stream;
                     ostringstream stream;
                     stream << js[id][field][i];
                     stream << js[id][field][i];
                     string pdb = trim(stream.str());
                     string pdb = trim(stream.str());
-                    if (find_id_pattern(pdb, benchmark)) {
+                    string pdb_complete = find_id_pattern(pdb, benchmark);
+                    if (!(pdb_complete.empty())) {
                         pair<string, string> p;
                         pair<string, string> p;
-                        p.first = pdb;
+                        p.first = pdb_complete;
                         p.second = id;
                         p.second = id;
                         association.push_back(p);
                         association.push_back(p);
                     }
                     }
@@ -81,20 +84,97 @@ vector<pair<string, string>> find_id(const string& bibli, const string& benchmar
             }
             }
         }
         }
     }
     }
-    
     lib.close();
     lib.close();
     return association;
     return association;
 }
 }
 
 
+bool does_it_match(const string& result, const string& id_motif) {
+    ifstream f_res(result);
+    if (f_res.is_open()) {
+        string name;
+        string seq;
+        string struc;
+        string contacts;
+
+        getline(f_res, name);
+        getline(f_res, seq);
+        while (getline(f_res, struc)) {
+            string motif_json = "JSON" + id_motif + " +";
+            if(struc.find(motif_json, 0) != string::npos) {
+                return true;
+            }
+            motif_json = "JSON" + id_motif + "\n";
+            if(struc.find(motif_json, 0) != string::npos) {
+                return true;
+            }
+            getline(f_res,contacts);
+        }
+        f_res.close();
+    }
+    return false;
+}
+
+vector<string> select_not_motif(const string& bibli, const string& benchmark) {
+    vector<string> selection;
+    vector<pair<string, string>> association = find_id(bibli, benchmark);
+    vector<string> list_bm = get_list_pdb_benchmark(benchmark);
+
+    string path_begin = "/mnt/c/Users/natha/Documents/IBISC/biorseo2/biorseo/results/test_";
+    string path_MFE_F = ".json_pmF_MEA";
+
+    for (pair<string, string> p : association) {
+        string id_motif = p.second;
+        selection.push_back(id_motif);
+    }
+    for (pair<string, string> p : association) {
+        cout << p.first << ", " << p.second << endl;
+    }
+    cout << "size: " << association.size() << endl;
+   
+    for (string pdb : list_bm) {
+        string path_result = path_begin + pdb + path_MFE_F;
+        for (pair<string,string> pair : association) {
+            if (pair.first.substr(0, pair.first.size()-2).compare(pdb.substr(0, pdb.size()-2)) != 0) {
+                bool test = does_it_match(path_result, pair.second);
+
+                if (test) {
+                    //if (!(pair.second.compare("954"))) { cout << "p1: " << pair.first << "pdb: " << pdb << endl;}
+                    auto position = find(selection.begin(), selection.end(), pair.second);
+                    if (position != selection.end()) {
+                        int index = position - selection.begin();
+                        selection.erase(selection.begin() + index);
+                    }
+                }
+            }
+        }
+    }
+    sort(selection.begin(), selection.end() );
+    selection.erase(unique(selection.begin(), selection.end() ), selection.end() );
+
+    cout << "size: " << selection.size() << endl;
+
+    return selection;
+}
+
 int main()
 int main()
 {
 {
     string bibli = "/mnt/c/Users/natha/Documents/IBISC/biorseo2/biorseo/data/modules/ISAURE/Motifs_derniere_version/motifs_final.json";
     string bibli = "/mnt/c/Users/natha/Documents/IBISC/biorseo2/biorseo/data/modules/ISAURE/Motifs_derniere_version/motifs_final.json";
     string benchmark = "/mnt/c/Users/natha/Documents/IBISC/biorseo2/biorseo/data/modules/ISAURE/Motifs_version_initiale/benchmark.dbn";
     string benchmark = "/mnt/c/Users/natha/Documents/IBISC/biorseo2/biorseo/data/modules/ISAURE/Motifs_version_initiale/benchmark.dbn";
 
 
-    vector<pair<string, string>> association = find_id(bibli, benchmark);
-    /*for (pair<string,string> p : association) {
+    /*vector<pair<string, string>> association = find_id(bibli, benchmark);
+    for (pair<string,string> p : association) {
         cout << "<" << p.first << ", " << p.second << ">" << endl;
         cout << "<" << p.first << ", " << p.second << ">" << endl;
     }*/
     }*/
 
 
+    vector<string> selection = select_not_motif(bibli, benchmark);
+    for (string str : selection) {
+        cout << str << ", ";
+    }
+    cout << endl;
+
+    /*string result = "/mnt/c/Users/natha/Documents/IBISC/biorseo2/biorseo/results/test_1U6P_B.json_pmF_MEA";
+    bool test = does_it_match(result, "150");
+    cout << "test : " << test << endl;*/
+
     return 0;
     return 0;
 }
 }
\ No newline at end of file