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;
  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                                 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 = 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                                         #ifdef USE_MPI
  59                                                 int pid;
  60                                                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
  61                                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
  62
  63                                                 if (pid == 0) {
  64                                         #endif
  65
  66                                         ableToOpen = openInputFile(fastaFileNames[i], in);
  67                                         in.close();
  68
  69                                         #ifdef USE_MPI
  70                                                         for (int j = 1; j < processors; j++) {
  71                                                                 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
  72                                                         }
  73                                                 }else{
  74                                                         MPI_Status status;
  75                                                         MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
  76                                                 }
  77
  78                                         #endif
  79
  80                                         if (ableToOpen == 1) {
  81                                                 m->mothurOut(fastaFileNames[i] + " will be disregarded."); m->mothurOutEndLine();
  82                                                 //erase from file list
  83                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
  84                                                 i--;
  85                                         }
  86
  87                                 }
  88
  89                                 //make sure there is at least one valid file left
  90                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
  91                         }
  92
  93                         //if the user changes the output directory command factory will send this info to us in the output parameter
  94                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  95                                 outputDir = "";
  96                                 outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
  97                         }
  98
  99                         string temp;
 100                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 101                         filter = isTrue(temp);
 102
 103                         temp = validParameter.validFile(parameters, "correction", false);               if (temp == "not found") { temp = "T"; }
 104                         correction = isTrue(temp);
 105
 106                         temp = validParameter.validFile(parameters, "processors", false);               if (temp == "not found") { temp = "1"; }
 107                         convert(temp, processors);
 108
 109                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 110                         convert(temp, window);
 111
 112                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "25"; }
 113                         convert(temp, increment);
 114                 }
 115         }
 116         catch(exception& e) {
 117                 m->errorOut(e, "ChimeraBellerophonCommand", "ChimeraBellerophonCommand");
 118                 exit(1);
 119         }
 120 }
 121 //**********************************************************************************************************************
 122
 123 void ChimeraBellerophonCommand::help(){
 124         try {
 125                 m->mothurOut("The chimera.bellerophon command reads a fastafile and creates list of potentially chimeric sequences.\n");
 126                 m->mothurOut("The chimera.bellerophon command parameters are fasta, filter, correction, processors, window, increment. The fasta parameter is required.\n");
 127                 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");
 128                 m->mothurOut("The filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter, default=false. \n");
 129                 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");
 130                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 131                 #ifdef USE_MPI
 132                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 133                 #endif
 134                 m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras, default is 1/4 sequence length. \n");
 135                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 25.\n");
 136                 m->mothurOut("chimera.bellerophon(fasta=yourFastaFile, filter=yourFilter, correction=yourCorrection, processors=yourProcessors) \n");
 137                 m->mothurOut("Example: chimera.bellerophon(fasta=AD.align, filter=True, correction=true, window=200) \n");
 138                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 139         }
 140         catch(exception& e) {
 141                 m->errorOut(e, "ChimeraBellerophonCommand", "help");
 142                 exit(1);
 143         }
 144 }
 145
 146 //***************************************************************************************************************
 147
 148 ChimeraBellerophonCommand::~ChimeraBellerophonCommand(){        /*      do nothing      */      }
 149
 150 //***************************************************************************************************************
 151
 152 int ChimeraBellerophonCommand::execute(){
 153         try{
 154
 155                 if (abort == true) { return 0; }
 156
 157                 for (int i = 0; i < fastaFileNames.size(); i++) {
 158
 159                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 160
 161                         int start = time(NULL);
 162
 163                         chimera = new Bellerophon(fastaFileNames[i], filter, correction, window, increment, processors, outputDir);
 164
 165                         string outputFileName = outputDir + getRootName(getSimpleName(fastaFileNames[i])) +  "bellerophon.chimeras";
 166                         string accnosFileName = outputDir + getRootName(getSimpleName(fastaFileNames[i])) + "bellerophon.accnos";
 167                         bool hasAccnos = true;
 168
 169                         chimera->getChimeras();
 170
 171                         if (m->control_pressed) { delete chimera; for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0;     }
 172
 173                 #ifdef USE_MPI
 174                         MPI_File outMPI;
 175                         MPI_File outMPIAccnos;
 176
 177                         int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 178
 179                         char outFilename[1024];
 180                         strcpy(outFilename, accnosFileName.c_str());
 181
 182                         char FileName[1024];
 183                         strcpy(FileName, outputFileName.c_str());
 184
 185                         MPI_File_open(MPI_COMM_WORLD, FileName, outMode, MPI_INFO_NULL, &outMPI);  //comm, filename, mode, info, filepointer
 186                         MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 187
 188                         numSeqs = chimera->print(outMPI, outMPIAccnos);
 189
 190                         MPI_File_close(&outMPI);
 191                         MPI_File_close(&outMPIAccnos);
 192
 193                 #else
 194
 195                         ofstream out;
 196                         openOutputFile(outputFileName, out);
 197
 198                         ofstream out2;
 199                         openOutputFile(accnosFileName, out2);
 200
 201                         numSeqs = chimera->print(out, out2);
 202                         out.close();
 203                         out2.close();
 204
 205                 #endif
 206
 207                         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;       }
 208
 209                         //delete accnos file if its blank
 210                         if (isBlank(accnosFileName)) {  remove(accnosFileName.c_str());  hasAccnos = false; }
 211
 212                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 213
 214                         outputNames.push_back(outputFileName);
 215                         if (hasAccnos) {  outputNames.push_back(accnosFileName);  }
 216
 217                         delete chimera;
 218                 }
 219
 220                 m->mothurOutEndLine();
 221                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 222                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 223                 m->mothurOutEndLine();
 224
 225                 return 0;
 226
 227         }
 228         catch(exception& e) {
 229                 m->errorOut(e, "ChimeraBellerophonCommand", "execute");
 230                 exit(1);
 231         }
 232 }
 233 //**********************************************************************************************************************
 234