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 #ifdef __cplusplus
  12 extern "C" {
  13 #endif
  14 #include "clearcut.h"
  15 #ifdef __cplusplus
  16 }
  17 #endif
  18
  19 //**********************************************************************************************************************
  20 vector<string> ClearcutCommand::setParameters(){
  21         try {
  22                 CommandParameter pphylip("phylip", "InputTypes", "", "", "FastaPhylip", "FastaPhylip", "none",false,false); parameters.push_back(pphylip);
  23                 CommandParameter pfasta("fasta", "InputTypes", "", "", "FastaPhylip", "FastaPhylip", "none",false,false); parameters.push_back(pfasta);
  24                 CommandParameter pverbose("verbose", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pverbose);
  25                 CommandParameter pquiet("quiet", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pquiet);
  26                 CommandParameter pversion("version", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pversion);
  27                 CommandParameter pseed("seed", "String", "", "", "*", "", "",false,false); parameters.push_back(pseed);
  28                 CommandParameter pnorandom("norandom", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pnorandom);
  29                 CommandParameter pshuffle("shuffle", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pshuffle);
  30                 CommandParameter pneighbor("neighbor", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pneighbor);
  31                 CommandParameter pexpblen("expblen", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pexpblen);
  32                 CommandParameter pexpdist("expdist", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pexpdist);
  33                 CommandParameter pDNA("DNA", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pDNA);
  34                 CommandParameter pprotein("protein", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pprotein);
  35                 CommandParameter pjukes("jukes", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pjukes);
  36                 CommandParameter pkimura("kimura", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pkimura);
  37                 CommandParameter pstdout("stdout", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pstdout);
  38                 CommandParameter pntrees("ntrees", "Number", "", "1", "", "", "",false,false); parameters.push_back(pntrees);
  39                 CommandParameter pmatrixout("matrixout", "String", "", "", "", "", "",false,false); parameters.push_back(pmatrixout);
  40                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  41                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  42
  43                 vector<string> myArray;
  44                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  45                 return myArray;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "ClearcutCommand", "setParameters");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 string ClearcutCommand::getHelpString(){
  54         try {
  55                 string helpString = "";
  56                 helpString += "The clearcut command interfaces mothur with the clearcut program written by Initiative for Bioinformatics and Evolutionary Studies (IBEST) at the University of Idaho.\n";
  57                 helpString += "For more information about clearcut refer to http://bioinformatics.hungry.com/clearcut/ \n";
  58                 helpString += "The clearcut command parameters are phylip, fasta, version, verbose, quiet, seed, norandom, shuffle, neighbor, expblen, expdist, ntrees, matrixout, stdout, kimura, jukes, protein, DNA. \n";
  59                 helpString += "The phylip parameter allows you to enter your phylip formatted distance matrix. \n";
  60                 helpString += "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";
  61
  62                 helpString += "The version parameter prints out the version of clearcut you are using, default=F. \n";
  63                 helpString += "The verbose parameter prints out more output from clearcut, default=F. \n";
  64                 helpString += "The quiet parameter turns on silent operation mode, default=F. \n";
  65                 helpString += "The seed parameter allows you to explicitly set the PRNG seed to a specific value. \n";
  66                 helpString += "The norandom parameter allows you to attempt joins deterministically, default=F. \n";
  67                 helpString += "The shuffle parameter allows you to randomly shuffle the distance matrix, default=F. \n";
  68                 helpString += "The neighbor parameter allows you to use traditional Neighbor-Joining algorithm, default=T. \n";
  69
  70                 helpString += "The DNA parameter allows you to indicate your fasta file contains DNA sequences, default=F. \n";
  71                 helpString += "The protein parameter allows you to indicate your fasta file contains protein sequences, default=F. \n";
  72
  73                 helpString += "The stdout parameter outputs your tree to STDOUT, default=F. \n";
  74                 helpString += "The matrixout parameter allows you to specify a filename to output a distance matrix to. \n";
  75                 helpString += "The ntrees parameter allows you to specify the number of output trees, default=1. \n";
  76                 helpString += "The expblen parameter allows you to use exponential notation for branch lengths, default=F. \n";
  77                 helpString += "The expdist parameter allows you to use exponential notation for distance outputs, default=F. \n";
  78
  79                 helpString += "The clearcut command should be in the following format: \n";
  80                 helpString += "clearcut(phylip=yourDistanceFile) \n";
  81                 helpString += "Example: clearcut(phylip=abrecovery.phylip.dist) \n";
  82                 return helpString;
  83         }
  84         catch(exception& e) {
  85                 m->errorOut(e, "ClearcutCommand", "getHelpString");
  86                 exit(1);
  87         }
  88 }
  89
  90 //**********************************************************************************************************************
  91 ClearcutCommand::ClearcutCommand(){
  92         try {
  93                 abort = true; calledHelp = true;
  94                 setParameters();
  95                 vector<string> tempOutNames;
  96                 outputTypes["tree"] = tempOutNames;
  97                 outputTypes["matrixout"] = tempOutNames;
  98         }
  99         catch(exception& e) {
 100                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
 101                 exit(1);
 102         }
 103 }
 104 /**************************************************************************************/
 105 ClearcutCommand::ClearcutCommand(string option)  {
 106         try {
 107                 abort = false; calledHelp = false;
 108
 109                 //allow user to run help
 110                 if(option == "help") { help(); abort = true; calledHelp = true; }
 111                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 112
 113                 else {
 114                         vector<string> myArray = setParameters();
 115
 116                         OptionParser parser(option);
 117                         map<string,string> parameters = parser.getParameters();
 118
 119                         ValidParameters validParameter;
 120                         map<string, string>::iterator it;
 121
 122                         //check to make sure all parameters are valid for command
 123                         for (it = parameters.begin(); it != parameters.end(); it++) {
 124                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 125                         }
 126
 127                         //initialize outputTypes
 128                         vector<string> tempOutNames;
 129                         outputTypes["tree"] = tempOutNames;
 130                         outputTypes["matrixout"] = tempOutNames;
 131
 132                         //if the user changes the input directory command factory will send this info to us in the output parameter
 133                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 134                         if (inputDir == "not found"){   inputDir = "";          }
 135                         else {
 136                                 string path;
 137                                 it = parameters.find("fasta");
 138                                 //user has given a template file
 139                                 if(it != parameters.end()){
 140                                         path = m->hasPath(it->second);
 141                                         //if the user has not given a path then, add inputdir. else leave path alone.
 142                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 143                                 }
 144
 145                                 it = parameters.find("phylip");
 146                                 //user has given a template file
 147                                 if(it != parameters.end()){
 148                                         path = m->hasPath(it->second);
 149                                         //if the user has not given a path then, add inputdir. else leave path alone.
 150                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 151                                 }
 152                         }
 153
 154                         //check for required parameters
 155                         fastafile = validParameter.validFile(parameters, "fasta", true);
 156                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 157                         else if (fastafile == "not found") { fastafile = ""; }
 158                         else { inputFile = fastafile;  m->setFastaFile(fastafile); }
 159
 160                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 161                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
 162                         else if (phylipfile == "not found") { phylipfile = ""; }
 163                         else { inputFile = phylipfile;  m->setPhylipFile(phylipfile); }
 164
 165                         if ((phylipfile == "") && (fastafile == "")) {
 166                                 //is there are current file available for either of these?
 167                                 //give priority to phylip, then fasta
 168                                 phylipfile = m->getPhylipFile();
 169                                 if (phylipfile != "") {  inputFile = phylipfile; m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 170                                 else {
 171                                         fastafile = m->getFastaFile();
 172                                         if (fastafile != "") { inputFile = fastafile;  m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 173                                         else {
 174                                                 m->mothurOut("No valid current files. You must provide a phylip or fasta file before you can use the clearcut command."); m->mothurOutEndLine();
 175                                                 abort = true;
 176                                         }
 177                                 }
 178                         }
 179                         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; }
 180
 181
 182                         //if the user changes the output directory command factory will send this info to us in the output parameter
 183                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(inputFile);      }
 184
 185                         string temp;
 186                         temp = validParameter.validFile(parameters, "version", false);          if (temp == "not found"){       temp = "F";                     }
 187                         version = m->isTrue(temp);
 188
 189                         temp = validParameter.validFile(parameters, "verbose", false);          if (temp == "not found"){       temp = "F";                     }
 190                         verbose = m->isTrue(temp);
 191
 192                         temp = validParameter.validFile(parameters, "quiet", false);            if (temp == "not found"){       temp = "F";                     }
 193                         quiet = m->isTrue(temp);
 194
 195                         seed = validParameter.validFile(parameters, "seed", false);                     if (seed == "not found"){       seed = "*";                     }
 196
 197                         temp = validParameter.validFile(parameters, "norandom", false);         if (temp == "not found"){       temp = "F";                     }
 198                         norandom = m->isTrue(temp);
 199
 200                         temp = validParameter.validFile(parameters, "shuffle", false);          if (temp == "not found"){       temp = "F";                     }
 201                         shuffle = m->isTrue(temp);
 202
 203                         temp = validParameter.validFile(parameters, "neighbor", false);         if (temp == "not found"){       temp = "T";                     }
 204                         neighbor = m->isTrue(temp);
 205
 206                         temp = validParameter.validFile(parameters, "DNA", false);                      if (temp == "not found"){       temp = "F";                     }
 207                         DNA = m->isTrue(temp);
 208
 209                         temp = validParameter.validFile(parameters, "protein", false);          if (temp == "not found"){       temp = "F";                     }
 210                         protein = m->isTrue(temp);
 211
 212                         temp = validParameter.validFile(parameters, "jukes", false);            if (temp == "not found"){       temp = "F";                     }
 213                         jukes = m->isTrue(temp);
 214
 215                         temp = validParameter.validFile(parameters, "kimura", false);           if (temp == "not found"){       temp = "F";                     }
 216                         kimura = m->isTrue(temp);
 217
 218                         temp = validParameter.validFile(parameters, "stdout", false);           if (temp == "not found"){       temp = "F";                     }
 219                         stdoutWanted = m->isTrue(temp);
 220
 221                         matrixout = validParameter.validFile(parameters, "matrixout", false);   if (matrixout == "not found"){  matrixout = "";         }
 222
 223                         ntrees = validParameter.validFile(parameters, "ntrees", false);         if (ntrees == "not found"){     ntrees = "1";           }
 224
 225                         temp = validParameter.validFile(parameters, "expblen", false);          if (temp == "not found"){       temp = "F";                     }
 226                         expblen = m->isTrue(temp);
 227
 228                         temp = validParameter.validFile(parameters, "expdist", false);          if (temp == "not found"){       temp = "F";                     }
 229                         expdist = m->isTrue(temp);
 230
 231                         if ((fastafile != "") && ((!DNA) && (!protein))) { m->mothurOut("You must specify the type of sequences you are using: DNA or protein"); m->mothurOutEndLine(); abort=true; }
 232                 }
 233
 234         }
 235         catch(exception& e) {
 236                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
 237                 exit(1);
 238         }
 239 }
 240 /**************************************************************************************/
 241 int ClearcutCommand::execute() {
 242         try {
 243
 244                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 245
 246                 //prepare filename
 247                 string outputName = outputDir + m->getRootName(m->getSimpleName(inputFile)) + "tre";
 248                 outputNames.push_back(outputName); outputTypes["tree"].push_back(outputName);
 249
 250                 vector<char*> cPara;
 251
 252                 char* tempClearcut = new char[8];
 253                 strcpy(tempClearcut, "clearcut");
 254                 cPara.push_back(tempClearcut);
 255
 256                 //you gave us a distance matrix
 257                 if (phylipfile != "") {  char* temp = new char[10];  strcpy(temp, "--distance");  cPara.push_back(temp);        }
 258
 259                 //you gave us a fastafile
 260                 if (fastafile != "") { char* temp = new char[11];  strcpy(temp, "--alignment");  cPara.push_back(temp);         }
 261
 262                 if (version)                    {  char* temp = new char[9];  strcpy(temp, "--version");  cPara.push_back(temp);        }
 263                 if (verbose)                    {  char* temp = new char[9];  strcpy(temp, "--verbose");  cPara.push_back(temp);        }
 264                 if (quiet)                              {  char* temp = new char[7];  strcpy(temp, "--quiet");  cPara.push_back(temp);  }
 265                 if (seed != "*")                {
 266                         string tempSeed = "--seed=" + seed;
 267                         char* temp = new char[tempSeed.length()];
 268                         strcpy(temp, tempSeed.c_str());
 269                         cPara.push_back(temp);
 270                 }
 271                 if (norandom)                   {  char* temp = new char[10];  strcpy(temp, "--norandom");  cPara.push_back(temp);      }
 272                 if (shuffle)                    {  char* temp = new char[9];  strcpy(temp, "--shuffle");  cPara.push_back(temp);        }
 273                 if (neighbor)                   {  char* temp = new char[10];  strcpy(temp, "--neighbor");  cPara.push_back(temp);      }
 274
 275                 string tempIn = "--in=" + inputFile;
 276                 char* tempI = new char[tempIn.length()];
 277                 strcpy(tempI, tempIn.c_str());
 278                 cPara.push_back(tempI);
 279
 280                 if (stdoutWanted)               {  char* temp = new char[8];  strcpy(temp, "--stdout");  cPara.push_back(temp); }
 281                 else{
 282                         string tempOut = "--out=" + outputName;
 283
 284                         char* temp = new char[tempOut.length()];
 285                         strcpy(temp, tempOut.c_str());
 286                         cPara.push_back(temp);
 287                 }
 288
 289                 if (DNA)                                {  char* temp = new char[5];  strcpy(temp, "--DNA");  cPara.push_back(temp);            }
 290                 if (protein)                    {  char* temp = new char[9];  strcpy(temp, "--protein");  cPara.push_back(temp);        }
 291                 if (jukes)                              {  char* temp = new char[7];  strcpy(temp, "--jukes");  cPara.push_back(temp);          }
 292                 if (kimura)                             { char* temp = new char[8];  strcpy(temp, "--kimura");  cPara.push_back(temp);          }
 293                 if (matrixout != "")    {
 294                         string tempMatrix =  "--matrixout=" + outputDir + matrixout;
 295                         char* temp = new char[tempMatrix.length()];
 296                         strcpy(temp, tempMatrix.c_str());
 297                         cPara.push_back(temp);
 298                         outputNames.push_back((outputDir + matrixout));
 299                         outputTypes["matrixout"].push_back((outputDir + matrixout));
 300                 }
 301
 302                 if (ntrees != "1")              {
 303                         string tempNtrees = "--ntrees=" + ntrees;
 304                         char* temp = new char[tempNtrees.length()];
 305                         strcpy(temp, tempNtrees.c_str());
 306                         cPara.push_back(temp);
 307                 }
 308
 309                 if (expblen)                    { char* temp = new char[9];  strcpy(temp, "--expblen");  cPara.push_back(temp);         }
 310                 if (expdist)                    { char* temp = new char[9];  strcpy(temp, "--expdist");  cPara.push_back(temp); }
 311
 312                 char** clearcutParameters;
 313                 clearcutParameters = new char*[cPara.size()];
 314                 for (int i = 0; i < cPara.size(); i++) {  clearcutParameters[i] = cPara[i];  }
 315                 int numArgs = cPara.size();
 316
 317                 clearcut_main(numArgs, clearcutParameters);
 318
 319                 //free memory
 320                 for(int i = 0; i < cPara.size(); i++)  {  delete[] cPara[i];  }
 321                 delete[] clearcutParameters;
 322
 323                 if (!stdoutWanted) {
 324
 325                         //set first tree file as new current treefile
 326                         string currentTree = "";
 327                         itTypes = outputTypes.find("tree");
 328                         if (itTypes != outputTypes.end()) {
 329                                 if ((itTypes->second).size() != 0) { currentTree = (itTypes->second)[0]; m->setTreeFile(currentTree); }
 330                         }
 331
 332                         m->mothurOutEndLine();
 333                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 334                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 335                         m->mothurOutEndLine();
 336                 }
 337
 338                 return 0;
 339         }
 340         catch(exception& e) {
 341                 m->errorOut(e, "ClearcutCommand", "execute");
 342                 exit(1);
 343         }
 344 }
 345 /**************************************************************************************/
 346
 347
 348
 349