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