chimerabellerophoncommand.cpp

   1 /*
   2  *  chimerabellerophoncommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 4/1/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimerabellerophoncommand.h"
  11 #include "bellerophon.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ChimeraBellerophonCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none","none","none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pfilter("filter", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pfilter);
  18                 CommandParameter pcorrection("correction", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pcorrection);
  19                 CommandParameter pwindow("window", "Number", "", "0", "", "", "",false,false); parameters.push_back(pwindow);
  20                 CommandParameter pincrement("increment", "Number", "", "25", "", "", "",false,false); parameters.push_back(pincrement);
  21                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  22                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  23                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  24
  25                 vector<string> myArray;
  26                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "ChimeraBellerophonCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string ChimeraBellerophonCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The chimera.bellerophon command reads a fastafile and creates list of potentially chimeric sequences.\n";
  39                 helpString += "The chimera.bellerophon command parameters are fasta, filter, correction, processors, window, increment. The fasta parameter is required, unless you have a valid current file.\n";
  40                 helpString += "The fasta parameter is required.  You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n";
  41                 helpString += "The filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter, default=false. \n";
  42                 helpString += "The correction parameter allows you to put more emphasis on the distance between highly similar sequences and less emphasis on the differences between remote homologs, default=true.\n";
  43                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  44 #ifdef USE_MPI
  45                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  46 #endif
  47                 helpString += "The window parameter allows you to specify the window size for searching for chimeras, default is 1/4 sequence length. \n";
  48                 helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 25.\n";
  49                 helpString += "chimera.bellerophon(fasta=yourFastaFile, filter=yourFilter, correction=yourCorrection, processors=yourProcessors) \n";
  50                 helpString += "Example: chimera.bellerophon(fasta=AD.align, filter=True, correction=true, window=200) \n";
  51                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  52                 return helpString;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "ChimeraBellerophonCommand", "getHelpString");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 string ChimeraBellerophonCommand::getOutputFileNameTag(string type, string inputName=""){
  61         try {
  62         string outputFileName = "";
  63                 map<string, vector<string> >::iterator it;
  64
  65         //is this a type this command creates
  66         it = outputTypes.find(type);
  67         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  68         else {
  69             if (type == "chimera") {  outputFileName =  "bellerophon.chimeras"; }
  70             else if (type == "accnos") {  outputFileName =  "bellerophon.accnos"; }
  71             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  72         }
  73         return outputFileName;
  74         }
  75         catch(exception& e) {
  76                 m->errorOut(e, "ChimeraBellerophonCommand", "getOutputFileNameTag");
  77                 exit(1);
  78         }
  79 }
  80 //**********************************************************************************************************************
  81 ChimeraBellerophonCommand::ChimeraBellerophonCommand(){
  82         try {
  83                 abort = true; calledHelp = true;
  84                 setParameters();
  85                 vector<string> tempOutNames;
  86                 outputTypes["chimera"] = tempOutNames;
  87                 outputTypes["accnos"] = tempOutNames;
  88         }
  89         catch(exception& e) {
  90                 m->errorOut(e, "ChimeraBellerophonCommand", "ChimeraBellerophonCommand");
  91                 exit(1);
  92         }
  93 }
  94 //***************************************************************************************************************
  95 ChimeraBellerophonCommand::ChimeraBellerophonCommand(string option)  {
  96         try {
  97                 abort = false; calledHelp = false;
  98
  99                 //allow user to run help
 100                 if(option == "help") { help(); abort = true; calledHelp = true; }
 101                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 102
 103                 else {
 104
 105                         vector<string> myArray = setParameters();
 106
 107                         OptionParser parser(option);
 108                         map<string,string> parameters = parser.getParameters();
 109
 110                         ValidParameters validParameter("chimera.bellerophon");
 111                         map<string,string>::iterator it;
 112
 113                         //check to make sure all parameters are valid for command
 114                         for (it = parameters.begin(); it != parameters.end(); it++) {
 115                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 116                         }
 117
 118                         //initialize outputTypes
 119                         vector<string> tempOutNames;
 120                         outputTypes["chimera"] = tempOutNames;
 121                         outputTypes["accnos"] = tempOutNames;
 122
 123                         //if the user changes the input directory command factory will send this info to us in the output parameter
 124                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 125                         if (inputDir == "not found"){   inputDir = "";          }
 126
 127                         fastafile = validParameter.validFile(parameters, "fasta", false);
 128                         if (fastafile == "not found") {
 129                                 //if there is a current fasta file, use it
 130                                 string filename = m->getFastaFile();
 131                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 132                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 133                         }else {
 134                                 m->splitAtDash(fastafile, fastaFileNames);
 135
 136                                 //go through files and make sure they are good, if not, then disregard them
 137                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 138
 139                                         bool ignore = false;
 140                                         if (fastaFileNames[i] == "current") {
 141                                                 fastaFileNames[i] = m->getFastaFile();
 142                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 143                                                 else {
 144                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 145                                                         //erase from file list
 146                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 147                                                         i--;
 148                                                 }
 149                                         }
 150
 151                                         if (!ignore) {
 152
 153                                                 if (inputDir != "") {
 154                                                         string path = m->hasPath(fastaFileNames[i]);
 155                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 156                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 157                                                 }
 158
 159                                                 int ableToOpen;
 160                                                 ifstream in;
 161
 162                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 163
 164                                                 //if you can't open it, try default location
 165                                                 if (ableToOpen == 1) {
 166                                                         if (m->getDefaultPath() != "") { //default path is set
 167                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 168                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 169                                                                 ifstream in2;
 170                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 171                                                                 in2.close();
 172                                                                 fastaFileNames[i] = tryPath;
 173                                                         }
 174                                                 }
 175
 176                                                 //if you can't open it, try default location
 177                                                 if (ableToOpen == 1) {
 178                                                         if (m->getOutputDir() != "") { //default path is set
 179                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 180                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 181                                                                 ifstream in2;
 182                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 183                                                                 in2.close();
 184                                                                 fastaFileNames[i] = tryPath;
 185                                                         }
 186                                                 }
 187
 188                                                 in.close();
 189
 190                                                 if (ableToOpen == 1) {
 191                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 192                                                         //erase from file list
 193                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 194                                                         i--;
 195                                                 }else {
 196                                                         m->setFastaFile(fastaFileNames[i]);
 197                                                 }
 198                                         }
 199                                 }
 200
 201                                 //make sure there is at least one valid file left
 202                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 203                         }
 204
 205                         //if the user changes the output directory command factory will send this info to us in the output parameter
 206                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 207
 208                         string temp;
 209                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 210                         filter = m->isTrue(temp);
 211
 212                         temp = validParameter.validFile(parameters, "correction", false);               if (temp == "not found") { temp = "T"; }
 213                         correction = m->isTrue(temp);
 214
 215                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 216                         m->setProcessors(temp);
 217                         m->mothurConvert(temp, processors);
 218
 219                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 220                         m->mothurConvert(temp, window);
 221
 222                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "25"; }
 223                         m->mothurConvert(temp, increment);
 224                 }
 225         }
 226         catch(exception& e) {
 227                 m->errorOut(e, "ChimeraBellerophonCommand", "ChimeraBellerophonCommand");
 228                 exit(1);
 229         }
 230 }
 231 //***************************************************************************************************************
 232 int ChimeraBellerophonCommand::execute(){
 233         try{
 234
 235                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 236
 237                 for (int i = 0; i < fastaFileNames.size(); i++) {
 238
 239                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 240
 241                         int start = time(NULL);
 242
 243                         chimera = new Bellerophon(fastaFileNames[i], filter, correction, window, increment, processors, outputDir);
 244
 245                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 246                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) +  getOutputFileNameTag("chimera");
 247                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) +  getOutputFileNameTag("accnos");
 248
 249                         chimera->getChimeras();
 250
 251                         if (m->control_pressed) { delete chimera; for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear(); return 0;        }
 252
 253                 #ifdef USE_MPI
 254                         MPI_File outMPI;
 255                         MPI_File outMPIAccnos;
 256
 257                         int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 258
 259                         char outFilename[1024];
 260                         strcpy(outFilename, accnosFileName.c_str());
 261
 262                         char FileName[1024];
 263                         strcpy(FileName, outputFileName.c_str());
 264
 265                         MPI_File_open(MPI_COMM_WORLD, FileName, outMode, MPI_INFO_NULL, &outMPI);  //comm, filename, mode, info, filepointer
 266                         MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 267
 268                         numSeqs = chimera->print(outMPI, outMPIAccnos, "");
 269
 270                         MPI_File_close(&outMPI);
 271                         MPI_File_close(&outMPIAccnos);
 272
 273                 #else
 274
 275                         ofstream out;
 276                         m->openOutputFile(outputFileName, out);
 277
 278                         ofstream out2;
 279                         m->openOutputFile(accnosFileName, out2);
 280
 281                         numSeqs = chimera->print(out, out2, "");
 282                         out.close();
 283                         out2.close();
 284
 285                 #endif
 286
 287                         if (m->control_pressed) { m->mothurRemove(accnosFileName); m->mothurRemove(outputFileName); for (int i = 0; i < outputNames.size(); i++) {      m->mothurRemove(outputNames[i]);        } outputTypes.clear(); delete chimera;  return 0;       }
 288
 289                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 290
 291                         outputNames.push_back(outputFileName);  outputTypes["chimera"].push_back(outputFileName);
 292                         outputNames.push_back(accnosFileName);  outputTypes["accnos"].push_back(accnosFileName);
 293
 294                         delete chimera;
 295                 }
 296
 297                 //set accnos file as new current accnosfile
 298                 string current = "";
 299                 itTypes = outputTypes.find("accnos");
 300                 if (itTypes != outputTypes.end()) {
 301                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 302                 }
 303
 304                 m->mothurOutEndLine();
 305                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 306                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 307                 m->mothurOutEndLine();
 308
 309                 return 0;
 310
 311         }
 312         catch(exception& e) {
 313                 m->errorOut(e, "ChimeraBellerophonCommand", "execute");
 314                 exit(1);
 315         }
 316 }
 317 //**********************************************************************************************************************
 318