Full help in command line

@@ -19,6 +19,7 @@ bin/*
 
 
 # results
 # results
 results/*
 results/*
+benchmark_results/*
 log_of_the_run.sh
 log_of_the_run.sh
 logBadDesc.txt
 logBadDesc.txt
 gurobi.log
 gurobi.log

@@ -9,7 +9,7 @@ Contact : louis.becquey@univ-evry.fr
 1/ How it works
 1/ How it works
 ===================================
 ===================================
 INPUT:
 INPUT:
-- An RNA sequence (tested with sequences ~100 bases)
+- An RNA sequence (with 16 GB of RAM you can go up to ~230 bases)
 
 
 THEN
 THEN
 - **Pattern-matching step** : Find all possible occurrences of known RNAmodules in the query sequence, by finding subsequences of the querythat score well with the probabilistic models of the modules (like JAR3D, or BayesPairing)
 - **Pattern-matching step** : Find all possible occurrences of known RNAmodules in the query sequence, by finding subsequences of the querythat score well with the probabilistic models of the modules (like JAR3D, or BayesPairing)

@@ -21,19 +21,18 @@ bypdir = ""
 biorseoDir = "."
 biorseoDir = "."
 exec(compile(open(biorseoDir+"/EditMe").read(), '', 'exec'))
 exec(compile(open(biorseoDir+"/EditMe").read(), '', 'exec'))
 runDir = path.dirname(path.realpath(__file__))
 runDir = path.dirname(path.realpath(__file__))
-tempDir = biorseoDir + "/temp/"
-HLmotifDir = biorseoDir + "/data/modules/BGSU/HL/3.2/lib"
-ILmotifDir = biorseoDir + "/data/modules/BGSU/IL/3.2/lib"
-descfolder = biorseoDir + "/data/modules/DESC"
+modulespath = biorseoDir + "/data/modules"
+HLmotifDir = modulespath + "/BGSU/HL/3.2/lib"
+ILmotifDir = modulespath + "/BGSU/IL/3.2/lib"
+descfolder = modulespath + "/DESC"
 
 
 # Parse options
 # Parse options
 try:
 try:
-    opts, args = getopt.getopt(sys.argv[1:], "hi:o:", ["rna3dmotifs","3dmotifatlas","jar3d","bayespairing","patternmatch","func="])
-except getopt.GetoptError:
-    print("Please provide arguments !")
+    opts, args = getopt.getopt(sys.argv[1:], "bc:f:hi:jl:no:pt:v", ["verbose", "rna3dmotifs","3dmotifatlas","jar3d","bayespairing","patternmatch","func=","help","version","seq=","modules-path=", "first-objective=","output=","theta=","interrupt-limit="])
+except getopt.GetoptError as err:
+    print(err)
     sys.exit(2)
     sys.exit(2)
 
 
-
 m = Manager()
 m = Manager()
 running_stats = m.list()
 running_stats = m.list()
 running_stats.append(0) # n_launched
 running_stats.append(0) # n_launched
@@ -335,38 +334,75 @@ class BiorseoInstance:
         self.jobcount = 0
         self.jobcount = 0
         self.joblist = []
         self.joblist = []
         self.mode = 0 # default is single sequence mode
         self.mode = 0 # default is single sequence mode
+        self.forward_options = []
 
 
         for opt, arg in opts:
         for opt, arg in opts:
-            if opt == "-h":
-                print("biorseo.py -i myRNA.fa -o myRNA.rawB --rna3dmotifs --patternmatch --func B")
-                print("biorseo.py -i myRNA.fa -o myRNA.jar3dB --3dmotifatlas --jar3d --func B")
-                print("biorseo.py -i myRNA.fa -o myRNA.bgsubypD --3dmotifatlas --bayespairing --func D")
+            if opt == "-h" or opt == "--help":
+                print(  "Biorseo, Bi-Objective RNA Structure Efficient Optimizer\n"
+                        "Bio-objective integer linear programming framework to predict RNA secondary structures by including known RNA modules.\n"
+                        "Developped by Louis Becquey (louis.becquey@univ-evry.fr), 2019\n\n")
+                print("Usage:\tYou must provide:\n\t1) a FASTA input file with -i,\n\t2) a module type with --rna3dmotifs or --3dmotifatlas"
+                      "\n\t3) one module placement method in { --patternmatch, --jar3d, --bayespairing }\n\t")
+                print("Options:")
+                print("-h [ --help ]\t\tPrint this help message")
+                print("--version\t\t\tPrint the program version")
+                print("-i [ --seq ]\t\tFASTA file with the query RNA sequence")
+                print("-p [ --patternmatch ]\t\tUse regular expressions to place modules in the sequence")
+                print("-j [ --jar3d ]\t\tUse JAR3D to place modules in the sequence (requires --3dmotifatlas)")
+                print("-b [ --bayespairing ]\t\tUse BayesPairing to place modules in the sequence")
+                print("-o [ --output ]\t\tFolder where to output files")
+                print("-f [ --func ]\t\t(A, B, C or D, default is B)"
+                      "\t\t\t\tObjective function to score module insertions: (A) insert big modules (B) insert light, high-order modules"
+                      "\t\t\t\t(c) insert modules which score well with the sequence (D) insert light, high-order modules which score well with the sequence.")
+                      "\t\t\t\tC and D require cannot be used with --patternmatch.")
+
+                print("biorseo.py -i myRNA.fa -o myResultsFolder/ --rna3dmotifs --patternmatch --func B")
+                print("biorseo.py -i myRNA.fa -o myResultsFolder/ --3dmotifatlas --jar3d --func B")
+                print("biorseo.py -i myRNA.fa --3dmotifatlas --bayespairing --func D")
                 sys.exit()
                 sys.exit()
-            elif opt == "-i":
+            elif opt == "-i" or opt == "--seq":
                 self.inputfile = arg
                 self.inputfile = arg
-            elif opt == "-o":
+            elif opt == "-o" or opt == "--output":
                 self.outputf = arg # output file or folder...
                 self.outputf = arg # output file or folder...
                 if self.outputf[1] != '/':
                 if self.outputf[1] != '/':
                     self.outputf = getcwd() + '/' + self.outputf
                     self.outputf = getcwd() + '/' + self.outputf
                 if self.outputf[-1] != '/':
                 if self.outputf[-1] != '/':
                     self.outputf = self.outputf + '/'
                     self.outputf = self.outputf + '/'
-            elif opt == "--func":
+            elif opt == "-f" or opt == "--func":
                 if arg in ['A', 'B', 'C', 'D']:
                 if arg in ['A', 'B', 'C', 'D']:
                     self.func = arg
                     self.func = arg
                 else:
                 else:
                     raise "Unknown scoring function " + arg
                     raise "Unknown scoring function " + arg
-            elif opt == "--patternmatch":
+            elif opt == "-p" or opt == "--patternmatch":
                 self.type = "dpm"
                 self.type = "dpm"
-            elif opt == "--jar3d":
+            elif opt == "-j" or opt == "--jar3d":
                 self.type = "jar3d"
                 self.type = "jar3d"
-            elif opt == "--bayespairing":
+            elif opt == "-b" or opt == "--bayespairing":
                 self.type = "byp"
                 self.type = "byp"
             elif opt == "--rna3dmotifs":
             elif opt == "--rna3dmotifs":
                 self.modules = "desc"
                 self.modules = "desc"
             elif opt == "--3dmotifatlas":
             elif opt == "--3dmotifatlas":
                 self.modules = "bgsu"
                 self.modules = "bgsu"
-            else:
-                raise "Unknown option " + opt
+            elif opt == "--modulespath":
+                HLmotifDir = arg + "/HL/3.2/lib"
+                ILmotifDir = arg + "/IL/3.2/lib"
+                descfolder = arg
+            elif opt == "--version":
+                subprocess.call([biorseoDir+"/bin/biorseo", "--version"])
+                exit(0)
+            elif opt == "-l" or opt == "--interrupt-limit":
+                self.forward_options.append("-l")
+                self.forward_options.append(arg)
+            elif opt == "-v" or opt == "--verbose":
+                self.forward_options.append("-v")
+            elif opt == "-n" or opt == "--disable-pseudoknots":
+                self.forward_options.append("-n")
+            elif opt == "-t" or opt == "--theta":
+                self.forward_options.append("-t")
+                self.forward_options.append(arg)
+            elif opt == "-c" or opt == "--first-objective":
+                self.forward_options.append("-c")
+                self.forward_options.append(arg)
 
 
         print("saving files to", self.outputf)
         print("saving files to", self.outputf)
         # create jobs
         # create jobs
@@ -793,7 +829,8 @@ class BiorseoInstance:
             command = [executable, "-s", fastafile ]
             command = [executable, "-s", fastafile ]
             if method_type:
             if method_type:
                 command += [ method_type, csv ]
                 command += [ method_type, csv ]
-            command += [ "-o", self.outputf + instance.header + ext + self.func, "--type", self.func ]
+            command += [ "-o", self.outputf + instance.header + ext + self.func, "--function", self.func ]
+            command += self.forward_options
             self.joblist.append(Job(command=command, priority=priority, timeout=3600, how_many_in_parallel=3))
             self.joblist.append(Job(command=command, priority=priority, timeout=3600, how_many_in_parallel=3))
             
             
 
 

@@ -21,6 +21,7 @@ char   MOIP::obj_function_nbr_ = 'A';
 uint   MOIP::obj_to_solve_     = 1;
 uint   MOIP::obj_to_solve_     = 1;
 double MOIP::precision_        = 1e-5;
 double MOIP::precision_        = 1e-5;
 bool   MOIP::allow_pk_         = true;
 bool   MOIP::allow_pk_         = true;
+uint   MOIP::max_sol_nbr_      = 500;
 
 
 unsigned getNumConstraints(IloModel& m)
 unsigned getNumConstraints(IloModel& m)
 {
 {
@@ -499,8 +500,8 @@ void MOIP::add_solution(const SecondaryStructure& s)
 {
 {
     if (verbose_) cout << "\t>adding structure to Pareto set :\t" << s.to_string() << endl;
     if (verbose_) cout << "\t>adding structure to Pareto set :\t" << s.to_string() << endl;
     pareto_.push_back(s);
     pareto_.push_back(s);
-    if (pareto_.size() > 500) {
-        cerr << "\033[31m Quitting because combinatorial issues (>500 solutions in Pareto set). \033[0m" << endl;
+    if (pareto_.size() > max_sol_nbr_) {
+        cerr << "\033[31m Quitting because combinatorial issues (>" << max_sol_nbr_ << " solutions in Pareto set). \033[0m" << endl;
         exit(1);
         exit(1);
     }
     }
 }
 }

@@ -30,7 +30,8 @@ class MOIP
     static uint               obj_to_solve_;  // What objective do you prefer to solve in mono-objective portions of the algorithm ?
     static uint               obj_to_solve_;  // What objective do you prefer to solve in mono-objective portions of the algorithm ?
     static double             precision_;   // decimals to keep in objective values, to avoid numerical issues. otherwise, solution with objective 5.0000000009 dominates solution with 5.0 =(
     static double             precision_;   // decimals to keep in objective values, to avoid numerical issues. otherwise, solution with objective 5.0000000009 dominates solution with 5.0 =(
     static bool               allow_pk_;      // Wether we forbid pseudoknots (false) or allow them (true)
     static bool               allow_pk_;      // Wether we forbid pseudoknots (false) or allow them (true)
-
+    static uint               max_sol_nbr_;  // Number of solutions to accept in the Pareto set before we give up the computation
+    
     private:
     private:
     bool   is_undominated_yet(const SecondaryStructure& s);
     bool   is_undominated_yet(const SecondaryStructure& s);
     void   define_problem_constraints(void);
     void   define_problem_constraints(void);

@@ -73,14 +73,15 @@ int main(int argc, char* argv[])
     ("version", "Print the program version")
     ("version", "Print the program version")
     ("seq,s", po::value<string>(&inputName)->required(), "Fasta file containing the RNA sequence")
     ("seq,s", po::value<string>(&inputName)->required(), "Fasta file containing the RNA sequence")
     ("descfolder,d", po::value<string>(&motifs_path_name), "A folder containing modules in .desc format, as produced by Djelloul & Denise's catalog program")
     ("descfolder,d", po::value<string>(&motifs_path_name), "A folder containing modules in .desc format, as produced by Djelloul & Denise's catalog program")
-    ("jar3dcsv", po::value<string>(&motifs_path_name), "A file containing the output of JAR3D's search for motifs in the sequence, as produced by test_on_RNAstrand.py")
-    ("bayespaircsv", po::value<string>(&motifs_path_name), "A file containing the output of BayesPairing's search for motifs in the sequence, as produced by test_on_RNAstrand.py")
+    ("jar3dcsv,j", po::value<string>(&motifs_path_name), "A file containing the output of JAR3D's search for motifs in the sequence, as produced by test_on_RNAstrand.py")
+    ("bayespaircsv,b", po::value<string>(&motifs_path_name), "A file containing the output of BayesPairing's search for motifs in the sequence, as produced by 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")
     ("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")
     ("output,o", po::value<string>(&outputName), "A file to summarize the computation results")
     ("output,o", po::value<string>(&outputName), "A file to summarize the computation results")
     ("theta,t", po::value<float>(&theta_p_threshold)->default_value(0.001), "Pairing probability threshold to consider or not the possibility of pairing")
     ("theta,t", po::value<float>(&theta_p_threshold)->default_value(0.001), "Pairing probability threshold to consider or not the possibility of pairing")
-    ("type,f", po::value<char>(&obj_function_nbr)->default_value('A'), "What objective function to use to include motifs: square of motif size in nucleotides like "
-    "RNA-MoIP (A), motif size + number of components (B), site score (C), motif size + site score + number of components (D)")
+    ("function,f", po::value<char>(&obj_function_nbr)->default_value('B'), "What objective function to use to include motifs: square of motif size in nucleotides like "
+    "RNA-MoIP (A), light motif size + high number of components (B), site score (C), light motif size + site score + high number of components (D)")
     ("disable-pseudoknots,n", "Add constraints forbidding the formation of pseudoknots")
     ("disable-pseudoknots,n", "Add constraints forbidding the formation of pseudoknots")
+    ("limit,l", po::value<unsigned int>(&MOIP::max_sol_nbr_)->default_value(500), "Intermediate number of solutions in the Pareto set above which we give up the calculation.")
     ("verbose,v", "Print what is happening to stdout");
     ("verbose,v", "Print what is happening to stdout");
     po::variables_map vm;
     po::variables_map vm;
     po::store(po::parse_command_line(argc, argv, desc), vm);
     po::store(po::parse_command_line(argc, argv, desc), vm);
@@ -99,7 +100,7 @@ int main(int argc, char* argv[])
             return EXIT_SUCCESS;
             return EXIT_SUCCESS;
         }
         }
         if (vm.count("version")) {
         if (vm.count("version")) {
-            cout << "Biorseo v1.0, May 2019" << endl;
+            cout << "Biorseo v1.01, June 2019" << endl;
             return EXIT_SUCCESS;
             return EXIT_SUCCESS;
         }
         }
         if (vm.count("verbose")) verbose = true;
         if (vm.count("verbose")) verbose = true;
@@ -112,7 +113,7 @@ int main(int argc, char* argv[])
             return EXIT_FAILURE;
             return EXIT_FAILURE;
         }
         }
         if (vm.count("-d") and (obj_function_nbr == 'C' or obj_function_nbr == 'D')) {
         if (vm.count("-d") and (obj_function_nbr == 'C' or obj_function_nbr == 'D')) {
-            cerr << "\033[31mYou must provide --jar3dcsv or --bayespaircsv to use --type C or --type D.\033[0m See "
+            cerr << "\033[31mYou must provide --jar3dcsv or --bayespaircsv to use --function C or --function D.\033[0m See "
                     "--help for more "
                     "--help for more "
                     "information."
                     "information."
                  << endl;
                  << endl;