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