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