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","tree",false,false,true); parameters.push_back(pphylip);
  23                 CommandParameter pfasta("fasta", "InputTypes", "", "", "FastaPhylip", "FastaPhylip", "none","tree",false,false,true); 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::getOutputPattern(string type) {
  91     try {
  92         string pattern = "";
  93
  94         if (type == "matrixout") {  pattern = "[filename],"; }
  95         else if (type == "tree") {  pattern = "[filename],tre"; }
  96         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  97
  98         return pattern;
  99     }
 100     catch(exception& e) {
 101         m->errorOut(e, "ClearcutCommand", "getOutputPattern");
 102         exit(1);
 103     }
 104 }
 105 //**********************************************************************************************************************
 106 ClearcutCommand::ClearcutCommand(){
 107         try {
 108                 abort = true; calledHelp = true;
 109                 setParameters();
 110                 vector<string> tempOutNames;
 111                 outputTypes["tree"] = tempOutNames;
 112                 outputTypes["matrixout"] = tempOutNames;
 113         }
 114         catch(exception& e) {
 115                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
 116                 exit(1);
 117         }
 118 }
 119 /**************************************************************************************/
 120 ClearcutCommand::ClearcutCommand(string option)  {
 121         try {
 122                 abort = false; calledHelp = false;
 123
 124                 //allow user to run help
 125                 if(option == "help") { help(); abort = true; calledHelp = true; }
 126                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 127
 128                 else {
 129                         vector<string> myArray = setParameters();
 130
 131                         OptionParser parser(option);
 132                         map<string,string> parameters = parser.getParameters();
 133
 134                         ValidParameters validParameter;
 135                         map<string, string>::iterator it;
 136
 137                         //check to make sure all parameters are valid for command
 138                         for (it = parameters.begin(); it != parameters.end(); it++) {
 139                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 140                         }
 141
 142                         //initialize outputTypes
 143                         vector<string> tempOutNames;
 144                         outputTypes["tree"] = tempOutNames;
 145                         outputTypes["matrixout"] = tempOutNames;
 146
 147                         //if the user changes the input directory command factory will send this info to us in the output parameter
 148                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 149                         if (inputDir == "not found"){   inputDir = "";          }
 150                         else {
 151                                 string path;
 152                                 it = parameters.find("fasta");
 153                                 //user has given a template file
 154                                 if(it != parameters.end()){
 155                                         path = m->hasPath(it->second);
 156                                         //if the user has not given a path then, add inputdir. else leave path alone.
 157                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 158                                 }
 159
 160                                 it = parameters.find("phylip");
 161                                 //user has given a template file
 162                                 if(it != parameters.end()){
 163                                         path = m->hasPath(it->second);
 164                                         //if the user has not given a path then, add inputdir. else leave path alone.
 165                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 166                                 }
 167                         }
 168
 169                         //check for required parameters
 170                         fastafile = validParameter.validFile(parameters, "fasta", true);
 171                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 172                         else if (fastafile == "not found") { fastafile = ""; }
 173                         else { inputFile = fastafile;  m->setFastaFile(fastafile); }
 174
 175                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 176                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
 177                         else if (phylipfile == "not found") { phylipfile = ""; }
 178                         else { inputFile = phylipfile;  m->setPhylipFile(phylipfile); }
 179
 180                         if ((phylipfile == "") && (fastafile == "")) {
 181                                 //is there are current file available for either of these?
 182                                 //give priority to phylip, then fasta
 183                                 phylipfile = m->getPhylipFile();
 184                                 if (phylipfile != "") {  inputFile = phylipfile; m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 185                                 else {
 186                                         fastafile = m->getFastaFile();
 187                                         if (fastafile != "") { inputFile = fastafile;  m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 188                                         else {
 189                                                 m->mothurOut("No valid current files. You must provide a phylip or fasta file before you can use the clearcut command."); m->mothurOutEndLine();
 190                                                 abort = true;
 191                                         }
 192                                 }
 193                         }
 194                         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; }
 195
 196
 197                         //if the user changes the output directory command factory will send this info to us in the output parameter
 198                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(inputFile);      }
 199
 200                         string temp;
 201                         temp = validParameter.validFile(parameters, "version", false);          if (temp == "not found"){       temp = "F";                     }
 202                         version = m->isTrue(temp);
 203
 204                         temp = validParameter.validFile(parameters, "verbose", false);          if (temp == "not found"){       temp = "F";                     }
 205                         verbose = m->isTrue(temp);
 206
 207                         temp = validParameter.validFile(parameters, "quiet", false);            if (temp == "not found"){       temp = "F";                     }
 208                         quiet = m->isTrue(temp);
 209
 210                         seed = validParameter.validFile(parameters, "seed", false);                     if (seed == "not found"){       seed = "*";                     }
 211
 212                         temp = validParameter.validFile(parameters, "norandom", false);         if (temp == "not found"){       temp = "F";                     }
 213                         norandom = m->isTrue(temp);
 214
 215                         temp = validParameter.validFile(parameters, "shuffle", false);          if (temp == "not found"){       temp = "F";                     }
 216                         shuffle = m->isTrue(temp);
 217
 218                         temp = validParameter.validFile(parameters, "neighbor", false);         if (temp == "not found"){       temp = "T";                     }
 219                         neighbor = m->isTrue(temp);
 220
 221                         temp = validParameter.validFile(parameters, "DNA", false);                      if (temp == "not found"){       temp = "F";                     }
 222                         DNA = m->isTrue(temp);
 223
 224                         temp = validParameter.validFile(parameters, "protein", false);          if (temp == "not found"){       temp = "F";                     }
 225                         protein = m->isTrue(temp);
 226
 227                         temp = validParameter.validFile(parameters, "jukes", false);            if (temp == "not found"){       temp = "F";                     }
 228                         jukes = m->isTrue(temp);
 229
 230                         temp = validParameter.validFile(parameters, "kimura", false);           if (temp == "not found"){       temp = "F";                     }
 231                         kimura = m->isTrue(temp);
 232
 233                         temp = validParameter.validFile(parameters, "stdout", false);           if (temp == "not found"){       temp = "F";                     }
 234                         stdoutWanted = m->isTrue(temp);
 235
 236                         matrixout = validParameter.validFile(parameters, "matrixout", false);   if (matrixout == "not found"){  matrixout = "";         }
 237
 238                         ntrees = validParameter.validFile(parameters, "ntrees", false);         if (ntrees == "not found"){     ntrees = "1";           }
 239
 240                         temp = validParameter.validFile(parameters, "expblen", false);          if (temp == "not found"){       temp = "F";                     }
 241                         expblen = m->isTrue(temp);
 242
 243                         temp = validParameter.validFile(parameters, "expdist", false);          if (temp == "not found"){       temp = "F";                     }
 244                         expdist = m->isTrue(temp);
 245
 246                         if ((fastafile != "") && ((!DNA) && (!protein))) { m->mothurOut("You must specify the type of sequences you are using: DNA or protein"); m->mothurOutEndLine(); abort=true; }
 247                 }
 248
 249         }
 250         catch(exception& e) {
 251                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
 252                 exit(1);
 253         }
 254 }
 255 /**************************************************************************************/
 256 int ClearcutCommand::execute() {
 257         try {
 258
 259                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 260
 261                 //prepare filename
 262         map<string, string> variables;
 263         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputFile));
 264                 string outputName = getOutputFileName("tree", variables);
 265                 outputNames.push_back(outputName); outputTypes["tree"].push_back(outputName);
 266
 267                 vector<char*> cPara;
 268
 269                 char* tempClearcut = new char[9];
 270                 *tempClearcut = '\0'; strncat(tempClearcut, "clearcut", 8);
 271                 cPara.push_back(tempClearcut);
 272
 273                 //you gave us a distance matrix
 274                 if (phylipfile != "") {  char* temp = new char[11];  *temp = '\0'; strncat(temp, "--distance", 10);  cPara.push_back(temp);     }
 275
 276                 //you gave us a fastafile
 277                 if (fastafile != "") { char* temp = new char[12];  *temp = '\0'; strncat(temp, "--alignment", 11);  cPara.push_back(temp);      }
 278
 279                 if (version)                    {  char* temp = new char[10];  *temp = '\0'; strncat(temp, "--version", 9);  cPara.push_back(temp);     }
 280                 if (verbose)                    {  char* temp = new char[10];  *temp = '\0'; strncat(temp, "--verbose", 9);  cPara.push_back(temp);     }
 281                 if (quiet)                              {  char* temp = new char[8];  *temp = '\0'; strncat(temp, "--quiet", 7);  cPara.push_back(temp);        }
 282                 if (seed != "*")                {
 283                         string tempSeed = "--seed=" + seed;
 284                         char* temp = new char[tempSeed.length()+1];
 285                         *temp = '\0'; strncat(temp, tempSeed.c_str(), tempSeed.length());
 286                         cPara.push_back(temp);
 287                 }
 288                 if (norandom)                   {  char* temp = new char[11];  *temp = '\0'; strncat(temp, "--norandom", 10);  cPara.push_back(temp);   }
 289                 if (shuffle)                    {  char* temp = new char[10];  *temp = '\0'; strncat(temp, "--shuffle", 9);  cPara.push_back(temp);     }
 290                 if (neighbor)                   {  char* temp = new char[11];  *temp = '\0'; strncat(temp, "--neighbor", 10);  cPara.push_back(temp);   }
 291
 292                 string tempIn = "--in=" + inputFile;
 293                 char* tempI = new char[tempIn.length()+1];
 294                 *tempI = '\0'; strncat(tempI, tempIn.c_str(), tempIn.length());
 295                 cPara.push_back(tempI);
 296
 297                 if (stdoutWanted)               {  char* temp = new char[9];  *temp = '\0'; strncat(temp, "--stdout", 8);  cPara.push_back(temp);       }
 298                 else{
 299                         string tempOut = "--out=" + outputName;
 300                         char* temp = new char[tempOut.length()+1];
 301                         *temp = '\0'; strncat(temp, tempOut.c_str(), tempOut.length());
 302                         cPara.push_back(temp);
 303                 }
 304
 305                 if (DNA)                                {  char* temp = new char[6];  *temp = '\0'; strncat(temp, "--DNA", 5);  cPara.push_back(temp);          }
 306                 if (protein)                    {  char* temp = new char[10];  *temp = '\0'; strncat(temp, "--protein", 9);  cPara.push_back(temp);     }
 307                 if (jukes)                              {  char* temp = new char[8];  *temp = '\0'; strncat(temp, "--jukes", 7);  cPara.push_back(temp);                }
 308                 if (kimura)                             { char* temp = new char[9];  *temp = '\0'; strncat(temp, "--kimura", 8);  cPara.push_back(temp);                }
 309                 if (matrixout != "")    {
 310                         string tempMatrix =  "--matrixout=" + outputDir + matrixout;
 311                         char* temp = new char[tempMatrix.length()+1];
 312                         *temp = '\0'; strncat(temp, tempMatrix.c_str(), tempMatrix.length());
 313                         cPara.push_back(temp);
 314                         outputNames.push_back((outputDir + matrixout));
 315                         outputTypes["matrixout"].push_back((outputDir + matrixout));
 316                 }
 317
 318                 if (ntrees != "1")              {
 319                         string tempNtrees = "--ntrees=" + ntrees;
 320                         char* temp = new char[tempNtrees.length()+1];
 321                         *temp = '\0'; strncat(temp, tempNtrees.c_str(), tempNtrees.length());
 322                         cPara.push_back(temp);
 323                 }
 324
 325                 if (expblen)                    { char* temp = new char[10];  *temp = '\0'; strncat(temp, "--expblen", 9);  cPara.push_back(temp);      }
 326                 if (expdist)                    { char* temp = new char[10];  *temp = '\0'; strncat(temp, "--expdist", 9);  cPara.push_back(temp);      }
 327
 328                 char** clearcutParameters;
 329                 clearcutParameters = new char*[cPara.size()];
 330                 for (int i = 0; i < cPara.size(); i++) {  clearcutParameters[i] = cPara[i];  }
 331                 int numArgs = cPara.size();
 332
 333                 clearcut_main(numArgs, clearcutParameters);
 334
 335                 //free memory
 336                 for(int i = 0; i < cPara.size(); i++)  {  delete[] cPara[i];  }
 337                 delete[] clearcutParameters;
 338
 339                 if (!stdoutWanted) {
 340
 341                         //set first tree file as new current treefile
 342                         string currentTree = "";
 343                         itTypes = outputTypes.find("tree");
 344                         if (itTypes != outputTypes.end()) {
 345                                 if ((itTypes->second).size() != 0) { currentTree = (itTypes->second)[0]; m->setTreeFile(currentTree); }
 346                         }
 347
 348                         m->mothurOutEndLine();
 349                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 350                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 351                         m->mothurOutEndLine();
 352                 }
 353
 354                 return 0;
 355         }
 356         catch(exception& e) {
 357                 m->errorOut(e, "ClearcutCommand", "execute");
 358                 exit(1);
 359         }
 360 }
 361 /**************************************************************************************/
 362
 363
 364
 365