Louis BECQUEY / biorseo

Go to a project
Toggle navigation Toggle navigation pinning
Louis BECQUEY
c8d751c8 1 parent 2c032d9e

supports BayesPairing

Inline Side-by-side
Showing 3 changed files with 48 additions and 21 deletions
...@@ -193,14 +193,35 @@ void Motif::load_from_csv(string csv_line) ...@@ -193,14 +193,35 @@ void Motif::load_from_csv(string csv_line)
193 { 193 {
194 vector<string> tokens; 194 vector<string> tokens;
195 split(tokens, csv_line, boost::is_any_of(",")); 195 split(tokens, csv_line, boost::is_any_of(","));
196 - atlas_id = tokens[0]; 196 + if (csv_line.find(string("True")) != std::string::npos or csv_line.find(string("False")) != std::string::npos) { // This has been created by jar3d
197 - score_ = stoi(tokens[2]); 197 + atlas_id = tokens[0];
198 - comp.push_back(Component(make_pair<int, int>(stoi(tokens[3]), stoi(tokens[4])))); 198 + score_ = stoi(tokens[2]);
199 - if (tokens[5] != "-") comp.push_back(Component(make_pair<int, int>(stoi(tokens[5]), stoi(tokens[6])))); 199 + comp.push_back(Component(make_pair<int, int>(stoi(tokens[3]), stoi(tokens[4]))));
200 - reversed_ = (tokens[1] == "True"); 200 + if (tokens[5] != "-") comp.push_back(Component(make_pair<int, int>(stoi(tokens[5]), stoi(tokens[6]))));
201 - is_model_ = true; 201 + reversed_ = (tokens[1] == "True");
202 - PDBID = ""; 202 + is_model_ = true;
203 - source_ = RNAMOTIFATLAS; 203 + PDBID = "";
204 + source_ = RNAMOTIFATLAS;
205 + } else { // this has been created by BayesPairing
206 + score_ = stoi(tokens[1]);
207 + // identify source:
208 + if (tokens[0].find(string("rna3dmotif")) == std::string::npos) {
209 + is_model_ = true;
210 + PDBID = "";
211 + source_ = RNAMOTIFATLAS;
212 + atlas_id = tokens[0];
213 + } else {
214 + is_model_ = false;
215 + PDBID = tokens[0];
216 + source_ = RNA3DMOTIF;
217 + atlas_id = "";
218 + }
219 + uint i = 2;
220 + while (i < tokens.size()) {
221 + comp.push_back(Component(make_pair<int, int>(stoi(tokens[i]), stoi(tokens[i + 1]))));
222 + i += 2;
223 + }
224 + }
204 } 225 }
205 226
206 string Motif::pos_string(void) const 227 string Motif::pos_string(void) const
...@@ -378,7 +399,7 @@ vector<Motif> load_desc_folder(const string& path, const string& rna, bool verbo ...@@ -378,7 +399,7 @@ vector<Motif> load_desc_folder(const string& path, const string& rna, bool verbo
378 return posInsertionSites; 399 return posInsertionSites;
379 } 400 }
380 401
381 -vector<Motif> load_jar3d_output(const string& path) 402 +vector<Motif> load_csv(const string& path)
382 { 403 {
383 vector<Motif> posInsertionSites; 404 vector<Motif> posInsertionSites;
384 std::ifstream motifs; 405 std::ifstream motifs;
......
...@@ -41,7 +41,7 @@ class Motif ...@@ -41,7 +41,7 @@ class Motif
41 public: 41 public:
42 Motif(); 42 Motif();
43 Motif(const vector<Component>& v, string PDB); 43 Motif(const vector<Component>& v, string PDB);
44 - void load_from_csv(string csv_line); 44 + void load_from_csv(string csv_line);
45 // static void build_from_desc(path descfile, string rna, vector<Motif>& final_results); 45 // static void build_from_desc(path descfile, string rna, vector<Motif>& final_results);
46 static void build_from_desc(args_of_parallel_func args); 46 static void build_from_desc(args_of_parallel_func args);
47 static char is_valid_DESC(const string& descfile); 47 static char is_valid_DESC(const string& descfile);
...@@ -63,7 +63,7 @@ class Motif ...@@ -63,7 +63,7 @@ class Motif
63 63
64 bool is_desc_insertible(const string& descfile, const string& rna, bool verbose); 64 bool is_desc_insertible(const string& descfile, const string& rna, bool verbose);
65 vector<Motif> load_desc_folder(const string& path, const string& rna, bool verbose); 65 vector<Motif> load_desc_folder(const string& path, const string& rna, bool verbose);
66 -vector<Motif> load_jar3d_output(const string& path); 66 +vector<Motif> load_csv(const string& path);
67 67
68 // utilities to compare secondary structures: 68 // utilities to compare secondary structures:
69 bool operator==(const Motif& m1, const Motif& m2); 69 bool operator==(const Motif& m1, const Motif& m2);
......
...@@ -72,12 +72,14 @@ int main(int argc, char* argv[]) ...@@ -72,12 +72,14 @@ int main(int argc, char* argv[])
72 "descfolder,d", 72 "descfolder,d",
73 po::value<string>(&motifs_path_name), 73 po::value<string>(&motifs_path_name),
74 "A folder containing modules in .desc format, as produced by Djelloul & Denise's catalog program")( 74 "A folder containing modules in .desc format, as produced by Djelloul & Denise's catalog program")(
75 - "jar3dcsv,f", 75 + "jar3dcsv",
76 po::value<string>(&motifs_path_name), 76 po::value<string>(&motifs_path_name),
77 - "A file containing the output of JAR3D's search for motifs in the sequence, as produced by searchForMotifSites.py")( 77 + "A file containing the output of JAR3D's search for motifs in the sequence, as produced by test_on_RNAstrand.py")(
78 - "first-objective,c", 78 + "bayespaircsv",
79 - po::value<unsigned int>(&MOIP::obj_to_solve_)->default_value(1), 79 + po::value<string>(&motifs_path_name),
80 - "Objective to solve in the mono-objective portions of the algorithm")("output,o", po::value<string>(&outputName), "A file to summarize the computation results")( 80 + "A file containing the output of BayesPairing's search for motifs in the sequence, as produced by "
81 + "test_on_RNAstrand.py")("first-objective,c", po::value<unsigned int>(&MOIP::obj_to_solve_)->default_value(1), "Objective to solve in the mono-objective portions of the algorithm")(
82 + "output,o", po::value<string>(&outputName), "A file to summarize the computation results")(
81 "theta,t", 83 "theta,t",
82 po::value<float>(&theta_p_threshold)->default_value(0.001), 84 po::value<float>(&theta_p_threshold)->default_value(0.001),
83 "Pairing probability threshold to consider or not the possibility of pairing")( 85 "Pairing probability threshold to consider or not the possibility of pairing")(
...@@ -108,14 +110,16 @@ int main(int argc, char* argv[]) ...@@ -108,14 +110,16 @@ int main(int argc, char* argv[])
108 return EXIT_SUCCESS; 110 return EXIT_SUCCESS;
109 } 111 }
110 if (vm.count("verbose")) verbose = true; 112 if (vm.count("verbose")) verbose = true;
111 - if (!vm.count("jar3dcsv") and !vm.count("descfolder")) { 113 + if (!vm.count("jar3dcsv") and !vm.count("bayespaircsv") and !vm.count("descfolder")) {
112 - cerr << "\033[31mYou must provide at least --jar3dcsv or --descfolder.\033[0m See --help for more " 114 + cerr << "\033[31mYou must provide at least --jar3dcsv, --bayespaircsv or --descfolder.\033[0m See --help "
115 + "for more "
113 "information." 116 "information."
114 << endl; 117 << endl;
115 return EXIT_FAILURE; 118 return EXIT_FAILURE;
116 } 119 }
117 - if (vm.count("-d") and (obj_function_nbr == 1 or obj_function_nbr == 3)) { 120 + if (vm.count("-d") and (obj_function_nbr == 2 or obj_function_nbr == 3)) {
118 - cerr << "\033[31mYou must provide at -f to use --type 1 or --type 3.\033[0m See --help for more " 121 + cerr << "\033[31mYou must provide --jar3dcsv or --bayespaircsv to use --type 2 or --type 3.\033[0m See "
122 + "--help for more "
119 "information." 123 "information."
120 << endl; 124 << endl;
121 return EXIT_FAILURE; 125 return EXIT_FAILURE;
...@@ -150,7 +154,9 @@ int main(int argc, char* argv[]) ...@@ -150,7 +154,9 @@ int main(int argc, char* argv[])
150 return EXIT_FAILURE; 154 return EXIT_FAILURE;
151 } 155 }
152 if (vm.count("jar3dcsv")) 156 if (vm.count("jar3dcsv"))
153 - posInsertionSites = load_jar3d_output(motifs_path_name.c_str()); 157 + posInsertionSites = load_csv(motifs_path_name.c_str());
158 + else if (vm.count("bayespaircsv"))
159 + posInsertionSites = load_csv(motifs_path_name.c_str());
154 else 160 else
155 posInsertionSites = load_desc_folder(motifs_path_name.c_str(), fa->seq(), verbose); 161 posInsertionSites = load_desc_folder(motifs_path_name.c_str(), fa->seq(), verbose);
156 162
......