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