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 string ClearcutCommand::getOutputFileNameTag(string type, string inputName=""){
  91         try {
  92         string outputFileName = "";
  93                 map<string, vector<string> >::iterator it;
  94
  95         //is this a type this command creates
  96         it = outputTypes.find(type);
  97         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  98         else {
  99             if (type == "tree") {  outputFileName =  "tre"; }
 100             else if (type == "matrixout") {  outputFileName =  ""; }
 101             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
 102         }
 103         return outputFileName;
 104         }
 105         catch(exception& e) {
 106                 m->errorOut(e, "ClearcutCommand", "getOutputFileNameTag");
 107                 exit(1);
 108         }
 109 }
 110 //**********************************************************************************************************************
 111 ClearcutCommand::ClearcutCommand(){
 112         try {
 113                 abort = true; calledHelp = true;
 114                 setParameters();
 115                 vector<string> tempOutNames;
 116                 outputTypes["tree"] = tempOutNames;
 117                 outputTypes["matrixout"] = tempOutNames;
 118         }
 119         catch(exception& e) {
 120                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
 121                 exit(1);
 122         }
 123 }
 124 /**************************************************************************************/
 125 ClearcutCommand::ClearcutCommand(string option)  {
 126         try {
 127                 abort = false; calledHelp = false;
 128
 129                 //allow user to run help
 130                 if(option == "help") { help(); abort = true; calledHelp = true; }
 131                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 132
 133                 else {
 134                         vector<string> myArray = setParameters();
 135
 136                         OptionParser parser(option);
 137                         map<string,string> parameters = parser.getParameters();
 138
 139                         ValidParameters validParameter;
 140                         map<string, string>::iterator it;
 141
 142                         //check to make sure all parameters are valid for command
 143                         for (it = parameters.begin(); it != parameters.end(); it++) {
 144                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 145                         }
 146
 147                         //initialize outputTypes
 148                         vector<string> tempOutNames;
 149                         outputTypes["tree"] = tempOutNames;
 150                         outputTypes["matrixout"] = tempOutNames;
 151
 152                         //if the user changes the input directory command factory will send this info to us in the output parameter
 153                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 154                         if (inputDir == "not found"){   inputDir = "";          }
 155                         else {
 156                                 string path;
 157                                 it = parameters.find("fasta");
 158                                 //user has given a template file
 159                                 if(it != parameters.end()){
 160                                         path = m->hasPath(it->second);
 161                                         //if the user has not given a path then, add inputdir. else leave path alone.
 162                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 163                                 }
 164
 165                                 it = parameters.find("phylip");
 166                                 //user has given a template file
 167                                 if(it != parameters.end()){
 168                                         path = m->hasPath(it->second);
 169                                         //if the user has not given a path then, add inputdir. else leave path alone.
 170                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 171                                 }
 172                         }
 173
 174                         //check for required parameters
 175                         fastafile = validParameter.validFile(parameters, "fasta", true);
 176                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 177                         else if (fastafile == "not found") { fastafile = ""; }
 178                         else { inputFile = fastafile;  m->setFastaFile(fastafile); }
 179
 180                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 181                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
 182                         else if (phylipfile == "not found") { phylipfile = ""; }
 183                         else { inputFile = phylipfile;  m->setPhylipFile(phylipfile); }
 184
 185                         if ((phylipfile == "") && (fastafile == "")) {
 186                                 //is there are current file available for either of these?
 187                                 //give priority to phylip, then fasta
 188                                 phylipfile = m->getPhylipFile();
 189                                 if (phylipfile != "") {  inputFile = phylipfile; m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 190                                 else {
 191                                         fastafile = m->getFastaFile();
 192                                         if (fastafile != "") { inputFile = fastafile;  m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 193                                         else {
 194                                                 m->mothurOut("No valid current files. You must provide a phylip or fasta file before you can use the clearcut command."); m->mothurOutEndLine();
 195                                                 abort = true;
 196                                         }
 197                                 }
 198                         }
 199                         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; }
 200
 201
 202                         //if the user changes the output directory command factory will send this info to us in the output parameter
 203                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(inputFile);      }
 204
 205                         string temp;
 206                         temp = validParameter.validFile(parameters, "version", false);          if (temp == "not found"){       temp = "F";                     }
 207                         version = m->isTrue(temp);
 208
 209                         temp = validParameter.validFile(parameters, "verbose", false);          if (temp == "not found"){       temp = "F";                     }
 210                         verbose = m->isTrue(temp);
 211
 212                         temp = validParameter.validFile(parameters, "quiet", false);            if (temp == "not found"){       temp = "F";                     }
 213                         quiet = m->isTrue(temp);
 214
 215                         seed = validParameter.validFile(parameters, "seed", false);                     if (seed == "not found"){       seed = "*";                     }
 216
 217                         temp = validParameter.validFile(parameters, "norandom", false);         if (temp == "not found"){       temp = "F";                     }
 218                         norandom = m->isTrue(temp);
 219
 220                         temp = validParameter.validFile(parameters, "shuffle", false);          if (temp == "not found"){       temp = "F";                     }
 221                         shuffle = m->isTrue(temp);
 222
 223                         temp = validParameter.validFile(parameters, "neighbor", false);         if (temp == "not found"){       temp = "T";                     }
 224                         neighbor = m->isTrue(temp);
 225
 226                         temp = validParameter.validFile(parameters, "DNA", false);                      if (temp == "not found"){       temp = "F";                     }
 227                         DNA = m->isTrue(temp);
 228
 229                         temp = validParameter.validFile(parameters, "protein", false);          if (temp == "not found"){       temp = "F";                     }
 230                         protein = m->isTrue(temp);
 231
 232                         temp = validParameter.validFile(parameters, "jukes", false);            if (temp == "not found"){       temp = "F";                     }
 233                         jukes = m->isTrue(temp);
 234
 235                         temp = validParameter.validFile(parameters, "kimura", false);           if (temp == "not found"){       temp = "F";                     }
 236                         kimura = m->isTrue(temp);
 237
 238                         temp = validParameter.validFile(parameters, "stdout", false);           if (temp == "not found"){       temp = "F";                     }
 239                         stdoutWanted = m->isTrue(temp);
 240
 241                         matrixout = validParameter.validFile(parameters, "matrixout", false);   if (matrixout == "not found"){  matrixout = "";         }
 242
 243                         ntrees = validParameter.validFile(parameters, "ntrees", false);         if (ntrees == "not found"){     ntrees = "1";           }
 244
 245                         temp = validParameter.validFile(parameters, "expblen", false);          if (temp == "not found"){       temp = "F";                     }
 246                         expblen = m->isTrue(temp);
 247
 248                         temp = validParameter.validFile(parameters, "expdist", false);          if (temp == "not found"){       temp = "F";                     }
 249                         expdist = m->isTrue(temp);
 250
 251                         if ((fastafile != "") && ((!DNA) && (!protein))) { m->mothurOut("You must specify the type of sequences you are using: DNA or protein"); m->mothurOutEndLine(); abort=true; }
 252                 }
 253
 254         }
 255         catch(exception& e) {
 256                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
 257                 exit(1);
 258         }
 259 }
 260 /**************************************************************************************/
 261 int ClearcutCommand::execute() {
 262         try {
 263
 264                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 265
 266                 //prepare filename
 267                 string outputName = outputDir + m->getRootName(m->getSimpleName(inputFile)) + getOutputFileNameTag("tree");
 268                 outputNames.push_back(outputName); outputTypes["tree"].push_back(outputName);
 269
 270                 vector<char*> cPara;
 271
 272                 char* tempClearcut = new char[8];
 273                 strcpy(tempClearcut, "clearcut");
 274                 cPara.push_back(tempClearcut);
 275
 276                 //you gave us a distance matrix
 277                 if (phylipfile != "") {  char* temp = new char[10];  strcpy(temp, "--distance");  cPara.push_back(temp);        }
 278
 279                 //you gave us a fastafile
 280                 if (fastafile != "") { char* temp = new char[11];  strcpy(temp, "--alignment");  cPara.push_back(temp);         }
 281
 282                 if (version)                    {  char* temp = new char[9];  strcpy(temp, "--version");  cPara.push_back(temp);        }
 283                 if (verbose)                    {  char* temp = new char[9];  strcpy(temp, "--verbose");  cPara.push_back(temp);        }
 284                 if (quiet)                              {  char* temp = new char[7];  strcpy(temp, "--quiet");  cPara.push_back(temp);  }
 285                 if (seed != "*")                {
 286                         string tempSeed = "--seed=" + seed;
 287                         char* temp = new char[tempSeed.length()];
 288                         strcpy(temp, tempSeed.c_str());
 289                         cPara.push_back(temp);
 290                 }
 291                 if (norandom)                   {  char* temp = new char[10];  strcpy(temp, "--norandom");  cPara.push_back(temp);      }
 292                 if (shuffle)                    {  char* temp = new char[9];  strcpy(temp, "--shuffle");  cPara.push_back(temp);        }
 293                 if (neighbor)                   {  char* temp = new char[10];  strcpy(temp, "--neighbor");  cPara.push_back(temp);      }
 294
 295                 string tempIn = "--in=" + inputFile;
 296                 char* tempI = new char[tempIn.length()];
 297                 strcpy(tempI, tempIn.c_str());
 298                 cPara.push_back(tempI);
 299
 300                 if (stdoutWanted)               {  char* temp = new char[8];  strcpy(temp, "--stdout");  cPara.push_back(temp); }
 301                 else{
 302                         string tempOut = "--out=" + outputName;
 303
 304                         char* temp = new char[tempOut.length()];
 305                         strcpy(temp, tempOut.c_str());
 306                         cPara.push_back(temp);
 307                 }
 308
 309                 if (DNA)                                {  char* temp = new char[5];  strcpy(temp, "--DNA");  cPara.push_back(temp);            }
 310                 if (protein)                    {  char* temp = new char[9];  strcpy(temp, "--protein");  cPara.push_back(temp);        }
 311                 if (jukes)                              {  char* temp = new char[7];  strcpy(temp, "--jukes");  cPara.push_back(temp);          }
 312                 if (kimura)                             { char* temp = new char[8];  strcpy(temp, "--kimura");  cPara.push_back(temp);          }
 313                 if (matrixout != "")    {
 314                         string tempMatrix =  "--matrixout=" + outputDir + matrixout;
 315                         char* temp = new char[tempMatrix.length()];
 316                         strcpy(temp, tempMatrix.c_str());
 317                         cPara.push_back(temp);
 318                         outputNames.push_back((outputDir + matrixout));
 319                         outputTypes["matrixout"].push_back((outputDir + matrixout));
 320                 }
 321
 322                 if (ntrees != "1")              {
 323                         string tempNtrees = "--ntrees=" + ntrees;
 324                         char* temp = new char[tempNtrees.length()];
 325                         strcpy(temp, tempNtrees.c_str());
 326                         cPara.push_back(temp);
 327                 }
 328
 329                 if (expblen)                    { char* temp = new char[9];  strcpy(temp, "--expblen");  cPara.push_back(temp);         }
 330                 if (expdist)                    { char* temp = new char[9];  strcpy(temp, "--expdist");  cPara.push_back(temp); }
 331
 332                 char** clearcutParameters;
 333                 clearcutParameters = new char*[cPara.size()];
 334                 for (int i = 0; i < cPara.size(); i++) {  clearcutParameters[i] = cPara[i];  }
 335                 int numArgs = cPara.size();
 336
 337                 clearcut_main(numArgs, clearcutParameters);
 338
 339                 //free memory
 340                 for(int i = 0; i < cPara.size(); i++)  {  delete[] cPara[i];  }
 341                 delete[] clearcutParameters;
 342
 343                 if (!stdoutWanted) {
 344
 345                         //set first tree file as new current treefile
 346                         string currentTree = "";
 347                         itTypes = outputTypes.find("tree");
 348                         if (itTypes != outputTypes.end()) {
 349                                 if ((itTypes->second).size() != 0) { currentTree = (itTypes->second)[0]; m->setTreeFile(currentTree); }
 350                         }
 351
 352                         m->mothurOutEndLine();
 353                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 354                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 355                         m->mothurOutEndLine();
 356                 }
 357
 358                 return 0;
 359         }
 360         catch(exception& e) {
 361                 m->errorOut(e, "ClearcutCommand", "execute");
 362                 exit(1);
 363         }
 364 }
 365 /**************************************************************************************/
 366
 367
 368
 369