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","chimera-accnos",false,true,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,true); 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::getOutputPattern(string type) {
  61     try {
  62         string pattern = "";
  63
  64         if (type == "chimera") {  pattern = "[filename],bellerophon.chimeras"; }
  65         else if (type == "accnos") {  pattern = "[filename],bellerophon.accnos"; }
  66         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  67
  68         return pattern;
  69     }
  70     catch(exception& e) {
  71         m->errorOut(e, "ChimeraBellerophonCommand", "getOutputPattern");
  72         exit(1);
  73     }
  74 }
  75 //**********************************************************************************************************************
  76 ChimeraBellerophonCommand::ChimeraBellerophonCommand(){
  77         try {
  78                 abort = true; calledHelp = true;
  79                 setParameters();
  80                 vector<string> tempOutNames;
  81                 outputTypes["chimera"] = tempOutNames;
  82                 outputTypes["accnos"] = tempOutNames;
  83         }
  84         catch(exception& e) {
  85                 m->errorOut(e, "ChimeraBellerophonCommand", "ChimeraBellerophonCommand");
  86                 exit(1);
  87         }
  88 }
  89 //***************************************************************************************************************
  90 ChimeraBellerophonCommand::ChimeraBellerophonCommand(string option)  {
  91         try {
  92                 abort = false; calledHelp = false;
  93
  94                 //allow user to run help
  95                 if(option == "help") { help(); abort = true; calledHelp = true; }
  96                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  97
  98                 else {
  99
 100                         vector<string> myArray = setParameters();
 101
 102                         OptionParser parser(option);
 103                         map<string,string> parameters = parser.getParameters();
 104
 105                         ValidParameters validParameter("chimera.bellerophon");
 106                         map<string,string>::iterator it;
 107
 108                         //check to make sure all parameters are valid for command
 109                         for (it = parameters.begin(); it != parameters.end(); it++) {
 110                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 111                         }
 112
 113                         //initialize outputTypes
 114                         vector<string> tempOutNames;
 115                         outputTypes["chimera"] = tempOutNames;
 116                         outputTypes["accnos"] = tempOutNames;
 117
 118                         //if the user changes the input directory command factory will send this info to us in the output parameter
 119                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 120                         if (inputDir == "not found"){   inputDir = "";          }
 121
 122                         fastafile = validParameter.validFile(parameters, "fasta", false);
 123                         if (fastafile == "not found") {
 124                                 //if there is a current fasta file, use it
 125                                 string filename = m->getFastaFile();
 126                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 127                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 128                         }else {
 129                                 m->splitAtDash(fastafile, fastaFileNames);
 130
 131                                 //go through files and make sure they are good, if not, then disregard them
 132                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 133
 134                                         bool ignore = false;
 135                                         if (fastaFileNames[i] == "current") {
 136                                                 fastaFileNames[i] = m->getFastaFile();
 137                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 138                                                 else {
 139                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 140                                                         //erase from file list
 141                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 142                                                         i--;
 143                                                 }
 144                                         }
 145
 146                                         if (!ignore) {
 147
 148                                                 if (inputDir != "") {
 149                                                         string path = m->hasPath(fastaFileNames[i]);
 150                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 151                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 152                                                 }
 153
 154                                                 int ableToOpen;
 155                                                 ifstream in;
 156
 157                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 158
 159                                                 //if you can't open it, try default location
 160                                                 if (ableToOpen == 1) {
 161                                                         if (m->getDefaultPath() != "") { //default path is set
 162                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 163                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 164                                                                 ifstream in2;
 165                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 166                                                                 in2.close();
 167                                                                 fastaFileNames[i] = tryPath;
 168                                                         }
 169                                                 }
 170
 171                                                 //if you can't open it, try default location
 172                                                 if (ableToOpen == 1) {
 173                                                         if (m->getOutputDir() != "") { //default path is set
 174                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 175                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 176                                                                 ifstream in2;
 177                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 178                                                                 in2.close();
 179                                                                 fastaFileNames[i] = tryPath;
 180                                                         }
 181                                                 }
 182
 183                                                 in.close();
 184
 185                                                 if (ableToOpen == 1) {
 186                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 187                                                         //erase from file list
 188                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 189                                                         i--;
 190                                                 }else {
 191                                                         m->setFastaFile(fastaFileNames[i]);
 192                                                 }
 193                                         }
 194                                 }
 195
 196                                 //make sure there is at least one valid file left
 197                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 198                         }
 199
 200                         //if the user changes the output directory command factory will send this info to us in the output parameter
 201                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 202
 203                         string temp;
 204                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 205                         filter = m->isTrue(temp);
 206
 207                         temp = validParameter.validFile(parameters, "correction", false);               if (temp == "not found") { temp = "T"; }
 208                         correction = m->isTrue(temp);
 209
 210                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 211                         m->setProcessors(temp);
 212                         m->mothurConvert(temp, processors);
 213
 214                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 215                         m->mothurConvert(temp, window);
 216
 217                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "25"; }
 218                         m->mothurConvert(temp, increment);
 219                 }
 220         }
 221         catch(exception& e) {
 222                 m->errorOut(e, "ChimeraBellerophonCommand", "ChimeraBellerophonCommand");
 223                 exit(1);
 224         }
 225 }
 226 //***************************************************************************************************************
 227 int ChimeraBellerophonCommand::execute(){
 228         try{
 229
 230                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 231
 232                 for (int i = 0; i < fastaFileNames.size(); i++) {
 233
 234                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 235
 236                         int start = time(NULL);
 237
 238                         chimera = new Bellerophon(fastaFileNames[i], filter, correction, window, increment, processors, outputDir);
 239
 240                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 241             map<string, string> variables;
 242             variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i]));
 243                         string outputFileName = getOutputFileName("chimera", variables);
 244                         string accnosFileName = getOutputFileName("accnos", variables);
 245
 246                         chimera->getChimeras();
 247
 248                         if (m->control_pressed) { delete chimera; for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear(); return 0;        }
 249
 250                 #ifdef USE_MPI
 251                         MPI_File outMPI;
 252                         MPI_File outMPIAccnos;
 253
 254                         int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 255
 256                         char outFilename[1024];
 257                         strcpy(outFilename, accnosFileName.c_str());
 258
 259                         char FileName[1024];
 260                         strcpy(FileName, outputFileName.c_str());
 261
 262                         MPI_File_open(MPI_COMM_WORLD, FileName, outMode, MPI_INFO_NULL, &outMPI);  //comm, filename, mode, info, filepointer
 263                         MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 264
 265                         numSeqs = chimera->print(outMPI, outMPIAccnos, "");
 266
 267                         MPI_File_close(&outMPI);
 268                         MPI_File_close(&outMPIAccnos);
 269
 270                 #else
 271
 272                         ofstream out;
 273                         m->openOutputFile(outputFileName, out);
 274
 275                         ofstream out2;
 276                         m->openOutputFile(accnosFileName, out2);
 277
 278                         numSeqs = chimera->print(out, out2, "");
 279                         out.close();
 280                         out2.close();
 281
 282                 #endif
 283
 284                         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;       }
 285
 286                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 287
 288                         outputNames.push_back(outputFileName);  outputTypes["chimera"].push_back(outputFileName);
 289                         outputNames.push_back(accnosFileName);  outputTypes["accnos"].push_back(accnosFileName);
 290
 291                         delete chimera;
 292                 }
 293
 294                 //set accnos file as new current accnosfile
 295                 string current = "";
 296                 itTypes = outputTypes.find("accnos");
 297                 if (itTypes != outputTypes.end()) {
 298                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 299                 }
 300
 301                 m->mothurOutEndLine();
 302                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 303                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 304                 m->mothurOutEndLine();
 305
 306                 return 0;
 307
 308         }
 309         catch(exception& e) {
 310                 m->errorOut(e, "ChimeraBellerophonCommand", "execute");
 311                 exit(1);
 312         }
 313 }
 314 //**********************************************************************************************************************
 315