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