clearcutcommand.cpp

   1 /*
   2  *  clearcutcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 5/11/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "clearcutcommand.h"
  11 #include "clearcut.h"
  12
  13
  14 /**************************************************************************************/
  15 ClearcutCommand::ClearcutCommand(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","phylip","version","verbose","quiet","seed","norandom","shuffle","neighbor","expblen",
  25                                                                 "expdist","ntrees","matrixout","stdout","kimura","jukes","protein","DNA","outputdir","inputdir"};
  26                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string,string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32                         map<string, string>::iterator it;
  33
  34                         //check to make sure all parameters are valid for command
  35                         for (it = parameters.begin(); it != parameters.end(); it++) {
  36                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  37                         }
  38
  39                         //if the user changes the input directory command factory will send this info to us in the output parameter
  40                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  41                         if (inputDir == "not found"){   inputDir = "";          }
  42                         else {
  43                                 string path;
  44                                 it = parameters.find("fasta");
  45                                 //user has given a template file
  46                                 if(it != parameters.end()){
  47                                         path = m->hasPath(it->second);
  48                                         //if the user has not given a path then, add inputdir. else leave path alone.
  49                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  50                                 }
  51
  52                                 it = parameters.find("phylip");
  53                                 //user has given a template file
  54                                 if(it != parameters.end()){
  55                                         path = m->hasPath(it->second);
  56                                         //if the user has not given a path then, add inputdir. else leave path alone.
  57                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
  58                                 }
  59                         }
  60
  61                         //check for required parameters
  62                         fastafile = validParameter.validFile(parameters, "fasta", true);
  63                         if (fastafile == "not open") { fastafile = ""; abort = true; }
  64                         else if (fastafile == "not found") { fastafile = ""; }
  65                         else { inputFile = fastafile;  }
  66
  67                         phylipfile = validParameter.validFile(parameters, "phylip", true);
  68                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
  69                         else if (phylipfile == "not found") { phylipfile = ""; }
  70                         else { inputFile = phylipfile;  }
  71
  72                         if ((phylipfile == "") && (fastafile == "")) {  m->mothurOut("You must provide either a phylip formatted distance matrix or an aligned fasta file."); m->mothurOutEndLine(); abort=true; }
  73                         if ((phylipfile != "") && (fastafile != "")) {  m->mothurOut("You must provide either a phylip formatted distance matrix or an aligned fasta file, not BOTH."); m->mothurOutEndLine(); abort=true; }
  74
  75
  76                         //if the user changes the output directory command factory will send this info to us in the output parameter
  77                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(inputFile);      }
  78
  79                         string temp;
  80                         temp = validParameter.validFile(parameters, "version", false);          if (temp == "not found"){       temp = "F";                     }
  81                         version = m->isTrue(temp);
  82
  83                         temp = validParameter.validFile(parameters, "verbose", false);          if (temp == "not found"){       temp = "F";                     }
  84                         verbose = m->isTrue(temp);
  85
  86                         temp = validParameter.validFile(parameters, "quiet", false);            if (temp == "not found"){       temp = "F";                     }
  87                         quiet = m->isTrue(temp);
  88
  89                         seed = validParameter.validFile(parameters, "seed", false);                     if (seed == "not found"){       seed = "*";                     }
  90
  91                         temp = validParameter.validFile(parameters, "norandom", false);         if (temp == "not found"){       temp = "F";                     }
  92                         norandom = m->isTrue(temp);
  93
  94                         temp = validParameter.validFile(parameters, "shuffle", false);          if (temp == "not found"){       temp = "F";                     }
  95                         shuffle = m->isTrue(temp);
  96
  97                         temp = validParameter.validFile(parameters, "neighbor", false);         if (temp == "not found"){       temp = "T";                     }
  98                         neighbor = m->isTrue(temp);
  99
 100                         temp = validParameter.validFile(parameters, "DNA", false);                      if (temp == "not found"){       temp = "F";                     }
 101                         DNA = m->isTrue(temp);
 102
 103                         temp = validParameter.validFile(parameters, "protein", false);          if (temp == "not found"){       temp = "F";                     }
 104                         protein = m->isTrue(temp);
 105
 106                         temp = validParameter.validFile(parameters, "jukes", false);            if (temp == "not found"){       temp = "F";                     }
 107                         jukes = m->isTrue(temp);
 108
 109                         temp = validParameter.validFile(parameters, "kimura", false);           if (temp == "not found"){       temp = "F";                     }
 110                         kimura = m->isTrue(temp);
 111
 112                         temp = validParameter.validFile(parameters, "stdout", false);           if (temp == "not found"){       temp = "F";                     }
 113                         stdoutWanted = m->isTrue(temp);
 114
 115                         matrixout = validParameter.validFile(parameters, "matrixout", false);   if (matrixout == "not found"){  matrixout = "";         }
 116
 117                         ntrees = validParameter.validFile(parameters, "ntrees", false);         if (ntrees == "not found"){     ntrees = "1";           }
 118
 119                         temp = validParameter.validFile(parameters, "expblen", false);          if (temp == "not found"){       temp = "F";                     }
 120                         expblen = m->isTrue(temp);
 121
 122                         temp = validParameter.validFile(parameters, "expdist", false);          if (temp == "not found"){       temp = "F";                     }
 123                         expdist = m->isTrue(temp);
 124
 125                         if ((fastafile != "") && ((!DNA) && (!protein))) { m->mothurOut("You must specify the type of sequences you are using: DNA or protein"); m->mothurOutEndLine(); abort=true; }
 126                 }
 127
 128         }
 129         catch(exception& e) {
 130                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
 131                 exit(1);
 132         }
 133 }
 134 //**********************************************************************************************************************
 135
 136 void ClearcutCommand::help(){
 137         try {
 138                 m->mothurOut("The clearcut command interfaces mothur with the clearcut program written by Initiative for Bioinformatics and Evolutionary Studies (IBEST) at the University of Idaho.\n");
 139                 m->mothurOut("For more information about clearcut refer to http://bioinformatics.hungry.com/clearcut/ \n");
 140                 m->mothurOut("The clearcut command parameters are phylip, fasta, version, verbose, quiet, seed, norandom, shuffle, neighbor, expblen, expdist, ntrees, matrixout, stdout, kimura, jukes, protein, DNA. \n");
 141                 m->mothurOut("The phylip parameter allows you to enter your phylip formatted distance matrix. \n");
 142                 m->mothurOut("The fasta parameter allows you to enter your aligned fasta file, if you enter a fastafile you specify if the sequences are DNA or protein using the DNA or protein parameters. \n");
 143
 144                 m->mothurOut("The version parameter prints out the version of clearcut you are using, default=F. \n");
 145                 m->mothurOut("The verbose parameter prints out more output from clearcut, default=F. \n");
 146                 m->mothurOut("The quiet parameter turns on silent operation mode, default=F. \n");
 147                 m->mothurOut("The seed parameter allows you to explicitly set the PRNG seed to a specific value. \n");
 148                 m->mothurOut("The norandom parameter allows you to attempt joins deterministically, default=F. \n");
 149                 m->mothurOut("The shuffle parameter allows you to randomly shuffle the distance matrix, default=F. \n");
 150                 m->mothurOut("The neighbor parameter allows you to use traditional Neighbor-Joining algorithm, default=T. \n");
 151
 152                 m->mothurOut("The DNA parameter allows you to indicate your fasta file contains DNA sequences, default=F. \n");
 153                 m->mothurOut("The protein parameter allows you to indicate your fasta file contains protein sequences, default=F. \n");
 154
 155                 m->mothurOut("The stdout parameter outputs your tree to STDOUT, default=F. \n");
 156                 m->mothurOut("The matrixout parameter allows you to specify a filename to output a distance matrix to. \n");
 157                 m->mothurOut("The ntrees parameter allows you to specify the number of output trees, default=1. \n");
 158                 m->mothurOut("The expblen parameter allows you to use exponential notation for branch lengths, default=F. \n");
 159                 m->mothurOut("The expdist parameter allows you to use exponential notation for distance outputs, default=F. \n");
 160
 161                 m->mothurOut("The clearcut command should be in the following format: \n");
 162                 m->mothurOut("clearcut(phylip=yourDistanceFile) \n");
 163                 m->mothurOut("Example: clearcut(phylip=abrecovery.phylip.dist) \n");
 164
 165         }
 166         catch(exception& e) {
 167                 m->errorOut(e, "ClearcutCommand", "help");
 168                 exit(1);
 169         }
 170 }
 171
 172 /**************************************************************************************/
 173 int ClearcutCommand::execute() {
 174         try {
 175
 176                 if (abort == true) { return 0; }
 177
 178                 //prepare filename
 179                 string outputName = outputDir + m->getRootName(m->getSimpleName(inputFile)) + "tre";
 180
 181                 vector<char*> cPara;
 182
 183                 char* tempClearcut = new char[8];  strcpy(tempClearcut, "clearcut");  cPara.push_back(tempClearcut);
 184
 185                 //you gave us a distance matrix
 186                 if (phylipfile != "") {  char* temp = new char[10];  strcpy(temp, "--distance");  cPara.push_back(temp);        }
 187
 188                 //you gave us a fastafile
 189                 if (fastafile != "") { char* temp = new char[11];  strcpy(temp, "--alignment");  cPara.push_back(temp);         }
 190
 191                 if (version)                    {  char* temp = new char[9];  strcpy(temp, "--version");  cPara.push_back(temp);        }
 192                 if (verbose)                    {  char* temp = new char[9];  strcpy(temp, "--verbose");  cPara.push_back(temp);        }
 193                 if (quiet)                              {  char* temp = new char[7];  strcpy(temp, "--quiet");  cPara.push_back(temp);  }
 194                 if (seed != "*")                {
 195                         string tempSeed = "--seed=" + seed;
 196                         char* temp = new char[tempSeed.length()];
 197                         strcpy(temp, tempSeed.c_str());
 198                         cPara.push_back(temp);
 199                 }
 200                 if (norandom)                   {  char* temp = new char[10];  strcpy(temp, "--norandom");  cPara.push_back(temp);      }
 201                 if (shuffle)                    {  char* temp = new char[9];  strcpy(temp, "--shuffle");  cPara.push_back(temp);        }
 202                 if (neighbor)                   {  char* temp = new char[10];  strcpy(temp, "--neighbor");  cPara.push_back(temp);      }
 203
 204                 string tempIn = "";
 205                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 206                         tempIn += "--in=" + inputFile;
 207                 #else
 208                         tempIn += "--in=\"" + inputFile + "\"";
 209                 #endif
 210
 211                 char* tempI = new char[tempIn.length()];
 212                 strcpy(tempI, tempIn.c_str());
 213                 cPara.push_back(tempI);
 214
 215                 if (stdoutWanted)               {  char* temp = new char[8];  strcpy(temp, "--stdout");  cPara.push_back(temp); }
 216                 else{
 217                         string tempOut = "";
 218                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 219                                 tempOut += "--out=" + outputName;
 220                         #else
 221                                 tempOut += "--out=\"" + outputName + "\"";
 222                         #endif
 223
 224                         char* temp = new char[tempOut.length()];
 225                         strcpy(temp, tempOut.c_str());
 226                         cPara.push_back(temp);
 227                 }
 228
 229                 if (DNA)                                {  char* temp = new char[5];  strcpy(temp, "--DNA");  cPara.push_back(temp);            }
 230                 if (protein)                    {  char* temp = new char[9];  strcpy(temp, "--protein");  cPara.push_back(temp);        }
 231                 if (jukes)                              {  char* temp = new char[7];  strcpy(temp, "--jukes");  cPara.push_back(temp);          }
 232                 if (kimura)                             { char* temp = new char[8];  strcpy(temp, "--kimura");  cPara.push_back(temp);          }
 233                 if (matrixout != "")    {
 234                         string tempMatrix =  "--matrixout=" + outputDir + matrixout;
 235                         char* temp = new char[tempMatrix.length()];
 236                         strcpy(temp, tempMatrix.c_str());
 237                         cPara.push_back(temp);
 238                 }
 239
 240                 if (ntrees != "1")              {
 241                         string tempNtrees = "--ntrees=" + ntrees;
 242                         char* temp = new char[tempNtrees.length()];
 243                         strcpy(temp, tempNtrees.c_str());
 244                         cPara.push_back(temp);
 245                 }
 246
 247                 if (expblen)                    { char* temp = new char[9];  strcpy(temp, "--expblen");  cPara.push_back(temp);         }
 248                 if (expdist)                    { char* temp = new char[9];  strcpy(temp, "--expdist");  cPara.push_back(temp); }
 249
 250                 char** clearcutParameters;
 251                 clearcutParameters = new char*[cPara.size()];
 252                 for (int i = 0; i < cPara.size(); i++) {  clearcutParameters[i] = cPara[i];  }
 253                 int numArgs = cPara.size();
 254
 255                 clearcut_main(numArgs, clearcutParameters);
 256
 257                 //free memory
 258                 for(int i = 0; i < cPara.size(); i++)  {  delete[] cPara[i];  }
 259                 delete[] clearcutParameters;
 260
 261                 if (!stdoutWanted) {
 262                         m->mothurOutEndLine();
 263                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 264                         m->mothurOut(outputName); m->mothurOutEndLine();
 265                         if (matrixout != "")    {  m->mothurOut(outputDir+matrixout); m->mothurOutEndLine();  }
 266                         m->mothurOutEndLine();
 267                 }
 268
 269                 return 0;
 270         }
 271         catch(exception& e) {
 272                 m->errorOut(e, "ClearcutCommand", "execute");
 273                 exit(1);
 274         }
 275 }
 276 /**************************************************************************************/
 277
 278
 279
 280