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
  12 /**************************************************************************************/
  13 ClearcutCommand::ClearcutCommand(string option)  {
  14         try {
  15                 abort = false;
  16
  17                 //allow user to run help
  18                 if(option == "help") { help(); abort = true; }
  19
  20                 else {
  21                         //valid paramters for this command
  22                         string Array[] =  {"fasta","phylip","version","verbose","quiet","seed","norandom","shuffle","neighbor","expblen",
  23                                                                 "expdist","ntrees","matrixout","stdout","kimura","jukes","protein","DNA","outputdir","inputdir"};
  24                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  25
  26                         OptionParser parser(option);
  27                         map<string,string> parameters = parser.getParameters();
  28
  29                         ValidParameters validParameter;
  30                         map<string, string>::iterator it;
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         //if the user changes the input directory command factory will send this info to us in the output parameter
  38                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  39                         if (inputDir == "not found"){   inputDir = "";          }
  40                         else {
  41                                 string path;
  42                                 it = parameters.find("fasta");
  43                                 //user has given a template file
  44                                 if(it != parameters.end()){
  45                                         path = hasPath(it->second);
  46                                         //if the user has not given a path then, add inputdir. else leave path alone.
  47                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  48                                 }
  49
  50                                 it = parameters.find("phylip");
  51                                 //user has given a template file
  52                                 if(it != parameters.end()){
  53                                         path = hasPath(it->second);
  54                                         //if the user has not given a path then, add inputdir. else leave path alone.
  55                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
  56                                 }
  57                         }
  58
  59                         //check for required parameters
  60                         fastafile = validParameter.validFile(parameters, "fasta", true);
  61                         if (fastafile == "not open") { fastafile = ""; abort = true; }
  62                         else if (fastafile == "not found") { fastafile = ""; }
  63                         else { inputFile = fastafile;  }
  64
  65                         phylipfile = validParameter.validFile(parameters, "phylip", true);
  66                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
  67                         else if (phylipfile == "not found") { phylipfile = ""; }
  68                         else { inputFile = phylipfile;  }
  69
  70                         if ((phylipfile == "") && (fastafile == "")) {  m->mothurOut("You must provide either a phylip formatted distance matrix or an aligned fasta file."); m->mothurOutEndLine(); abort=true; }
  71                         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; }
  72
  73
  74                         //if the user changes the output directory command factory will send this info to us in the output parameter
  75                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
  76
  77                         string temp;
  78                         temp = validParameter.validFile(parameters, "version", false);          if (temp == "not found"){       temp = "F";                     }
  79                         version = isTrue(temp);
  80
  81                         temp = validParameter.validFile(parameters, "verbose", false);          if (temp == "not found"){       temp = "F";                     }
  82                         verbose = isTrue(temp);
  83
  84                         temp = validParameter.validFile(parameters, "quiet", false);            if (temp == "not found"){       temp = "F";                     }
  85                         quiet = isTrue(temp);
  86
  87                         seed = validParameter.validFile(parameters, "seed", false);                     if (seed == "not found"){       seed = "*";                     }
  88
  89                         temp = validParameter.validFile(parameters, "norandom", false);         if (temp == "not found"){       temp = "F";                     }
  90                         norandom = isTrue(temp);
  91
  92                         temp = validParameter.validFile(parameters, "shuffle", false);          if (temp == "not found"){       temp = "F";                     }
  93                         shuffle = isTrue(temp);
  94
  95                         temp = validParameter.validFile(parameters, "neighbor", false);         if (temp == "not found"){       temp = "F";                     }
  96                         neighbor = isTrue(temp);
  97
  98                         temp = validParameter.validFile(parameters, "DNA", false);                      if (temp == "not found"){       temp = "F";                     }
  99                         DNA = isTrue(temp);
 100
 101                         temp = validParameter.validFile(parameters, "protein", false);          if (temp == "not found"){       temp = "F";                     }
 102                         protein = isTrue(temp);
 103
 104                         temp = validParameter.validFile(parameters, "jukes", false);            if (temp == "not found"){       temp = "F";                     }
 105                         jukes = isTrue(temp);
 106
 107                         temp = validParameter.validFile(parameters, "kimura", false);           if (temp == "not found"){       temp = "F";                     }
 108                         kimura = isTrue(temp);
 109
 110                         temp = validParameter.validFile(parameters, "stdout", false);           if (temp == "not found"){       temp = "F";                     }
 111                         stdoutWanted = isTrue(temp);
 112
 113                         matrixout = validParameter.validFile(parameters, "matrixout", false);   if (matrixout == "not found"){  matrixout = "";         }
 114
 115                         ntrees = validParameter.validFile(parameters, "ntrees", false);         if (ntrees == "not found"){     ntrees = "1";           }
 116
 117                         temp = validParameter.validFile(parameters, "expblen", false);          if (temp == "not found"){       temp = "F";                     }
 118                         expblen = isTrue(temp);
 119
 120                         temp = validParameter.validFile(parameters, "expdist", false);          if (temp == "not found"){       temp = "F";                     }
 121                         expdist = isTrue(temp);
 122
 123                         if ((fastafile != "") && ((!DNA) && (!protein))) { m->mothurOut("You must specify the type of sequences you are using: DNA or protein"); m->mothurOutEndLine(); abort=true; }
 124                 }
 125
 126         }
 127         catch(exception& e) {
 128                 m->errorOut(e, "ClearcutCommand", "ClearcutCommand");
 129                 exit(1);
 130         }
 131 }
 132 //**********************************************************************************************************************
 133
 134 void ClearcutCommand::help(){
 135         try {
 136                 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");
 137                 m->mothurOut("For more information about clearcut refer to http://bioinformatics.hungry.com/clearcut/ \n");
 138                 m->mothurOut("The clearcut executable must be in a folder called clearcut in the same folder as your mothur executable, similar to mothur's requirements for using blast. \n");
 139                 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");
 140                 m->mothurOut("The phylip parameter allows you to enter your phylip formatted distance matrix. \n");
 141                 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");
 142
 143                 m->mothurOut("The version parameter prints out the version of clearcut you are using, default=F. \n");
 144                 m->mothurOut("The verbose parameter prints out more output from clearcut, default=F. \n");
 145                 m->mothurOut("The quiet parameter turns on silent operation mode, default=F. \n");
 146                 m->mothurOut("The seed parameter allows you to explicitly set the PRNG seed to a specific value. \n");
 147                 m->mothurOut("The norandom parameter allows you to attempt joins deterministically, default=F. \n");
 148                 m->mothurOut("The shuffle parameter allows you to randomly shuffle the distance matrix, default=F. \n");
 149                 m->mothurOut("The neighbor parameter allows you to use traditional Neighbor-Joining algorithm, default=F. \n");
 150
 151                 m->mothurOut("The DNA parameter allows you to indicate your fasta file contains DNA sequences, default=F. \n");
 152                 m->mothurOut("The protein parameter allows you to indicate your fasta file contains protein sequences, default=F. \n");
 153
 154                 m->mothurOut("The stdout parameter outputs your tree to STDOUT, default=F. \n");
 155                 m->mothurOut("The matrixout parameter allows you to specify a filename to output a distance matrix to. \n");
 156                 m->mothurOut("The ntrees parameter allows you to specify the number of output trees, default=1. \n");
 157                 m->mothurOut("The expblen parameter allows you to use exponential notation for branch lengths, default=F. \n");
 158                 m->mothurOut("The expdist parameter allows you to use exponential notation for distance outputs, default=F. \n");
 159
 160                 m->mothurOut("The clearcut command should be in the following format: \n");
 161                 m->mothurOut("clearcut(phylip=yourDistanceFile) \n");
 162                 m->mothurOut("Example: clearcut(phylip=abrecovery.phylip.dist) \n");
 163
 164         }
 165         catch(exception& e) {
 166                 m->errorOut(e, "ClearcutCommand", "help");
 167                 exit(1);
 168         }
 169 }
 170
 171 /**************************************************************************************/
 172 int ClearcutCommand::execute() {
 173         try {
 174
 175                 if (abort == true) { return 0; }
 176
 177                 //prepare filename
 178                 string outputName = outputDir + getRootName(getSimpleName(inputFile)) + "tre";
 179
 180                 //get location of clearcut
 181                 GlobalData* globaldata = GlobalData::getInstance();
 182                 string path = globaldata->argv;
 183                 path = path.substr(0, (path.find_last_of('m')));
 184
 185                 string clearcutCommand = "";
 186                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 187                         clearcutCommand = path + "clearcut/clearcut ";
 188                 #else
 189                         clearcutCommand = path + "clearcut\\clearcut ";
 190                 #endif
 191
 192                 //you gave us a distance matrix
 193                 if (phylipfile != "") { clearcutCommand += "--distance ";       }
 194
 195                 //you gave us a fastafile
 196                 if (fastafile != "") { clearcutCommand += "--alignment ";       }
 197
 198                 if (version)                    {  clearcutCommand += "--version ";             }
 199                 if (verbose)                    {  clearcutCommand += "--verbose ";             }
 200                 if (quiet)                              {  clearcutCommand += "--quiet ";               }
 201                 if (seed != "*")                {  clearcutCommand += "--seed=" + seed + " "; }
 202                 if (norandom)                   {  clearcutCommand += "--norandom ";    }
 203                 if (shuffle)                    {  clearcutCommand += "--shuffle ";             }
 204                 if (neighbor)                   {  clearcutCommand += "--neighbor ";    }
 205
 206                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 207                         clearcutCommand += "--in=" + inputFile + " ";
 208                 #else
 209                         clearcutCommand += "--in=\"" + inputFile + "\" ";
 210                 #endif
 211
 212                 if (stdoutWanted)               {  clearcutCommand += "--stdout ";              }
 213                 else{
 214                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 215                                 clearcutCommand += "--out=" + outputName + " "; }
 216                         #else
 217                                 clearcutCommand += "--out=\"" + outputName + "\" "; }
 218                         #endif
 219
 220                 if (DNA)                                {  clearcutCommand += "--DNA ";                 }
 221                 if (protein)                    {  clearcutCommand += "--protein ";             }
 222                 if (jukes)                              {  clearcutCommand += "--jukes ";               }
 223                 if (kimura)                             {  clearcutCommand += "--kimura ";              }
 224                 if (matrixout != "")    {  clearcutCommand += "--matrixout=" + matrixout + " ";  }
 225                 if (ntrees != "1")              {  clearcutCommand += "--ntrees=" + ntrees + " "; }
 226                 if (expblen)                    {  clearcutCommand += "--expblen ";             }
 227                 if (expdist)                    {  clearcutCommand += "--expdist ";             }
 228
 229                 //run clearcut
 230                 system(clearcutCommand.c_str());
 231
 232                 if (!stdoutWanted) {
 233                         m->mothurOutEndLine();
 234                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 235                         m->mothurOut(outputName); m->mothurOutEndLine();
 236                         if (matrixout != "")    {  m->mothurOut(matrixout); m->mothurOutEndLine();  }
 237                         m->mothurOutEndLine();
 238                 }
 239
 240                 return 0;
 241         }
 242         catch(exception& e) {
 243                 m->errorOut(e, "ClearcutCommand", "execute");
 244                 exit(1);
 245         }
 246 }
 247 /**************************************************************************************/
 248
 249
 250
 251