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 vector<string> ClearcutCommand::getValidParameters(){
  15         try {
  16                 string AlignArray[] =  {"fasta","phylip","version","verbose","quiet","seed","norandom","shuffle","neighbor","expblen",
  17                                                                 "expdist","ntrees","matrixout","stdout","kimura","jukes","protein","DNA","outputdir","inputdir"};
  18                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "ClearcutCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 ClearcutCommand::ClearcutCommand(){
  28         try {
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["tree"] = tempOutNames;
  32                 outputTypes["matrixout"] = tempOutNames;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
  36                 exit(1);
  37         }
  38 }//**********************************************************************************************************************
  39 vector<string> ClearcutCommand::getRequiredParameters(){
  40         try {
  41                 string Array[] =  {"fasta","phylip","or"};
  42                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "ClearcutCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> ClearcutCommand::getRequiredFiles(){
  52         try {
  53                 vector<string> myArray;
  54                 return myArray;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "ClearcutCommand", "getRequiredFiles");
  58                 exit(1);
  59         }
  60 }
  61 /**************************************************************************************/
  62 ClearcutCommand::ClearcutCommand(string option)  {
  63         try {
  64                 abort = false;
  65
  66                 //allow user to run help
  67                 if(option == "help") { help(); abort = true; }
  68
  69                 else {
  70                         //valid paramters for this command
  71                         string Array[] =  {"fasta","phylip","version","verbose","quiet","seed","norandom","shuffle","neighbor","expblen",
  72                                                                 "expdist","ntrees","matrixout","stdout","kimura","jukes","protein","DNA","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;
  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["tree"] = tempOutNames;
  89                         outputTypes["matrixout"] = tempOutNames;
  90
  91                         //if the user changes the input directory command factory will send this info to us in the output parameter
  92                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  93                         if (inputDir == "not found"){   inputDir = "";          }
  94                         else {
  95                                 string path;
  96                                 it = parameters.find("fasta");
  97                                 //user has given a template file
  98                                 if(it != parameters.end()){
  99                                         path = m->hasPath(it->second);
 100                                         //if the user has not given a path then, add inputdir. else leave path alone.
 101                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 102                                 }
 103
 104                                 it = parameters.find("phylip");
 105                                 //user has given a template file
 106                                 if(it != parameters.end()){
 107                                         path = m->hasPath(it->second);
 108                                         //if the user has not given a path then, add inputdir. else leave path alone.
 109                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 110                                 }
 111                         }
 112
 113                         //check for required parameters
 114                         fastafile = validParameter.validFile(parameters, "fasta", true);
 115                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 116                         else if (fastafile == "not found") { fastafile = ""; }
 117                         else { inputFile = fastafile;  }
 118
 119                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 120                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
 121                         else if (phylipfile == "not found") { phylipfile = ""; }
 122                         else { inputFile = phylipfile;  }
 123
 124                         if ((phylipfile == "") && (fastafile == "")) {  m->mothurOut("You must provide either a phylip formatted distance matrix or an aligned fasta file."); m->mothurOutEndLine(); abort=true; }
 125                         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; }
 126
 127
 128                         //if the user changes the output directory command factory will send this info to us in the output parameter
 129                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(inputFile);      }
 130
 131                         string temp;
 132                         temp = validParameter.validFile(parameters, "version", false);          if (temp == "not found"){       temp = "F";                     }
 133                         version = m->isTrue(temp);
 134
 135                         temp = validParameter.validFile(parameters, "verbose", false);          if (temp == "not found"){       temp = "F";                     }
 136                         verbose = m->isTrue(temp);
 137
 138                         temp = validParameter.validFile(parameters, "quiet", false);            if (temp == "not found"){       temp = "F";                     }
 139                         quiet = m->isTrue(temp);
 140
 141                         seed = validParameter.validFile(parameters, "seed", false);                     if (seed == "not found"){       seed = "*";                     }
 142
 143                         temp = validParameter.validFile(parameters, "norandom", false);         if (temp == "not found"){       temp = "F";                     }
 144                         norandom = m->isTrue(temp);
 145
 146                         temp = validParameter.validFile(parameters, "shuffle", false);          if (temp == "not found"){       temp = "F";                     }
 147                         shuffle = m->isTrue(temp);
 148
 149                         temp = validParameter.validFile(parameters, "neighbor", false);         if (temp == "not found"){       temp = "T";                     }
 150                         neighbor = m->isTrue(temp);
 151
 152                         temp = validParameter.validFile(parameters, "DNA", false);                      if (temp == "not found"){       temp = "F";                     }
 153                         DNA = m->isTrue(temp);
 154
 155                         temp = validParameter.validFile(parameters, "protein", false);          if (temp == "not found"){       temp = "F";                     }
 156                         protein = m->isTrue(temp);
 157
 158                         temp = validParameter.validFile(parameters, "jukes", false);            if (temp == "not found"){       temp = "F";                     }
 159                         jukes = m->isTrue(temp);
 160
 161                         temp = validParameter.validFile(parameters, "kimura", false);           if (temp == "not found"){       temp = "F";                     }
 162                         kimura = m->isTrue(temp);
 163
 164                         temp = validParameter.validFile(parameters, "stdout", false);           if (temp == "not found"){       temp = "F";                     }
 165                         stdoutWanted = m->isTrue(temp);
 166
 167                         matrixout = validParameter.validFile(parameters, "matrixout", false);   if (matrixout == "not found"){  matrixout = "";         }
 168
 169                         ntrees = validParameter.validFile(parameters, "ntrees", false);         if (ntrees == "not found"){     ntrees = "1";           }
 170
 171                         temp = validParameter.validFile(parameters, "expblen", false);          if (temp == "not found"){       temp = "F";                     }
 172                         expblen = m->isTrue(temp);
 173
 174                         temp = validParameter.validFile(parameters, "expdist", false);          if (temp == "not found"){       temp = "F";                     }
 175                         expdist = m->isTrue(temp);
 176
 177                         if ((fastafile != "") && ((!DNA) && (!protein))) { m->mothurOut("You must specify the type of sequences you are using: DNA or protein"); m->mothurOutEndLine(); abort=true; }
 178                 }
 179
 180         }
 181         catch(exception& e) {
 182                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
 183                 exit(1);
 184         }
 185 }
 186 //**********************************************************************************************************************
 187
 188 void ClearcutCommand::help(){
 189         try {
 190                 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");
 191                 m->mothurOut("For more information about clearcut refer to http://bioinformatics.hungry.com/clearcut/ \n");
 192                 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");
 193                 m->mothurOut("The phylip parameter allows you to enter your phylip formatted distance matrix. \n");
 194                 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");
 195
 196                 m->mothurOut("The version parameter prints out the version of clearcut you are using, default=F. \n");
 197                 m->mothurOut("The verbose parameter prints out more output from clearcut, default=F. \n");
 198                 m->mothurOut("The quiet parameter turns on silent operation mode, default=F. \n");
 199                 m->mothurOut("The seed parameter allows you to explicitly set the PRNG seed to a specific value. \n");
 200                 m->mothurOut("The norandom parameter allows you to attempt joins deterministically, default=F. \n");
 201                 m->mothurOut("The shuffle parameter allows you to randomly shuffle the distance matrix, default=F. \n");
 202                 m->mothurOut("The neighbor parameter allows you to use traditional Neighbor-Joining algorithm, default=T. \n");
 203
 204                 m->mothurOut("The DNA parameter allows you to indicate your fasta file contains DNA sequences, default=F. \n");
 205                 m->mothurOut("The protein parameter allows you to indicate your fasta file contains protein sequences, default=F. \n");
 206
 207                 m->mothurOut("The stdout parameter outputs your tree to STDOUT, default=F. \n");
 208                 m->mothurOut("The matrixout parameter allows you to specify a filename to output a distance matrix to. \n");
 209                 m->mothurOut("The ntrees parameter allows you to specify the number of output trees, default=1. \n");
 210                 m->mothurOut("The expblen parameter allows you to use exponential notation for branch lengths, default=F. \n");
 211                 m->mothurOut("The expdist parameter allows you to use exponential notation for distance outputs, default=F. \n");
 212
 213                 m->mothurOut("The clearcut command should be in the following format: \n");
 214                 m->mothurOut("clearcut(phylip=yourDistanceFile) \n");
 215                 m->mothurOut("Example: clearcut(phylip=abrecovery.phylip.dist) \n");
 216
 217         }
 218         catch(exception& e) {
 219                 m->errorOut(e, "ClearcutCommand", "help");
 220                 exit(1);
 221         }
 222 }
 223
 224 /**************************************************************************************/
 225 int ClearcutCommand::execute() {
 226         try {
 227
 228                 if (abort == true) { return 0; }
 229
 230                 //prepare filename
 231                 string outputName = outputDir + m->getRootName(m->getSimpleName(inputFile)) + "tre";
 232                 outputNames.push_back(outputName); outputTypes["tree"].push_back(outputName);
 233
 234                 vector<char*> cPara;
 235
 236                 char* tempClearcut = new char[8];
 237                 strcpy(tempClearcut, "clearcut");
 238                 cPara.push_back(tempClearcut);
 239
 240                 //you gave us a distance matrix
 241                 if (phylipfile != "") {  char* temp = new char[10];  strcpy(temp, "--distance");  cPara.push_back(temp);        }
 242
 243                 //you gave us a fastafile
 244                 if (fastafile != "") { char* temp = new char[11];  strcpy(temp, "--alignment");  cPara.push_back(temp);         }
 245
 246                 if (version)                    {  char* temp = new char[9];  strcpy(temp, "--version");  cPara.push_back(temp);        }
 247                 if (verbose)                    {  char* temp = new char[9];  strcpy(temp, "--verbose");  cPara.push_back(temp);        }
 248                 if (quiet)                              {  char* temp = new char[7];  strcpy(temp, "--quiet");  cPara.push_back(temp);  }
 249                 if (seed != "*")                {
 250                         string tempSeed = "--seed=" + seed;
 251                         char* temp = new char[tempSeed.length()];
 252                         strcpy(temp, tempSeed.c_str());
 253                         cPara.push_back(temp);
 254                 }
 255                 if (norandom)                   {  char* temp = new char[10];  strcpy(temp, "--norandom");  cPara.push_back(temp);      }
 256                 if (shuffle)                    {  char* temp = new char[9];  strcpy(temp, "--shuffle");  cPara.push_back(temp);        }
 257                 if (neighbor)                   {  char* temp = new char[10];  strcpy(temp, "--neighbor");  cPara.push_back(temp);      }
 258
 259                 string tempIn = "--in=" + inputFile;
 260                 char* tempI = new char[tempIn.length()];
 261                 strcpy(tempI, tempIn.c_str());
 262                 cPara.push_back(tempI);
 263
 264                 if (stdoutWanted)               {  char* temp = new char[8];  strcpy(temp, "--stdout");  cPara.push_back(temp); }
 265                 else{
 266                         string tempOut = "--out=" + outputName;
 267
 268                         char* temp = new char[tempOut.length()];
 269                         strcpy(temp, tempOut.c_str());
 270                         cPara.push_back(temp);
 271                 }
 272
 273                 if (DNA)                                {  char* temp = new char[5];  strcpy(temp, "--DNA");  cPara.push_back(temp);            }
 274                 if (protein)                    {  char* temp = new char[9];  strcpy(temp, "--protein");  cPara.push_back(temp);        }
 275                 if (jukes)                              {  char* temp = new char[7];  strcpy(temp, "--jukes");  cPara.push_back(temp);          }
 276                 if (kimura)                             { char* temp = new char[8];  strcpy(temp, "--kimura");  cPara.push_back(temp);          }
 277                 if (matrixout != "")    {
 278                         string tempMatrix =  "--matrixout=" + outputDir + matrixout;
 279                         char* temp = new char[tempMatrix.length()];
 280                         strcpy(temp, tempMatrix.c_str());
 281                         cPara.push_back(temp);
 282                         outputNames.push_back((outputDir + matrixout));
 283                         outputTypes["matrixout"].push_back((outputDir + matrixout));
 284                 }
 285
 286                 if (ntrees != "1")              {
 287                         string tempNtrees = "--ntrees=" + ntrees;
 288                         char* temp = new char[tempNtrees.length()];
 289                         strcpy(temp, tempNtrees.c_str());
 290                         cPara.push_back(temp);
 291                 }
 292
 293                 if (expblen)                    { char* temp = new char[9];  strcpy(temp, "--expblen");  cPara.push_back(temp);         }
 294                 if (expdist)                    { char* temp = new char[9];  strcpy(temp, "--expdist");  cPara.push_back(temp); }
 295
 296                 char** clearcutParameters;
 297                 clearcutParameters = new char*[cPara.size()];
 298                 for (int i = 0; i < cPara.size(); i++) {  clearcutParameters[i] = cPara[i];  }
 299                 int numArgs = cPara.size();
 300
 301                 clearcut_main(numArgs, clearcutParameters);
 302
 303                 //free memory
 304                 for(int i = 0; i < cPara.size(); i++)  {  delete[] cPara[i];  }
 305                 delete[] clearcutParameters;
 306
 307                 if (!stdoutWanted) {
 308                         m->mothurOutEndLine();
 309                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 310                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 311                         m->mothurOutEndLine();
 312                 }
 313
 314                 return 0;
 315         }
 316         catch(exception& e) {
 317                 m->errorOut(e, "ClearcutCommand", "execute");
 318                 exit(1);
 319         }
 320 }
 321 /**************************************************************************************/
 322
 323
 324
 325