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