readtreecommand.cpp

   1 /*
   2  *  readtreecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 1/23/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "readtreecommand.h"
  11
  12 //**********************************************************************************************************************
  13 ReadTreeCommand::ReadTreeCommand(string option)  {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  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[] =  {"tree","group","name","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                         globaldata->newRead();
  38
  39                         //if the user changes the input directory command factory will send this info to us in the output parameter
  40                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  41                         if (inputDir == "not found"){   inputDir = "";          }
  42                         else {
  43                                 string path;
  44                                 it = parameters.find("tree");
  45                                 //user has given a template file
  46                                 if(it != parameters.end()){
  47                                         path = m->hasPath(it->second);
  48                                         //if the user has not given a path then, add inputdir. else leave path alone.
  49                                         if (path == "") {       parameters["tree"] = inputDir + it->second;             }
  50                                 }
  51
  52                                 it = parameters.find("group");
  53                                 //user has given a template file
  54                                 if(it != parameters.end()){
  55                                         path = m->hasPath(it->second);
  56                                         //if the user has not given a path then, add inputdir. else leave path alone.
  57                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
  58                                 }
  59
  60                                 it = parameters.find("name");
  61                                 //user has given a template file
  62                                 if(it != parameters.end()){
  63                                         path = m->hasPath(it->second);
  64                                         //if the user has not given a path then, add inputdir. else leave path alone.
  65                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
  66                                 }
  67
  68                         }
  69
  70
  71                         //check for required parameters
  72                         treefile = validParameter.validFile(parameters, "tree", true);
  73                         if (treefile == "not open") { abort = true; }
  74                         else if (treefile == "not found") { treefile = ""; m->mothurOut("tree is a required parameter for the read.tree command."); m->mothurOutEndLine(); abort = true;  }
  75                         else {  globaldata->setTreeFile(treefile);  globaldata->setFormat("tree");      }
  76
  77                         groupfile = validParameter.validFile(parameters, "group", true);
  78                         if (groupfile == "not open") { abort = true; }
  79                         else if (groupfile == "not found") { groupfile = ""; m->mothurOut("group is a required parameter for the read.tree command."); m->mothurOutEndLine(); abort = true;     }
  80                         else {
  81                                 globaldata->setGroupFile(groupfile);
  82                                 //read in group map info.
  83                                 treeMap = new TreeMap(groupfile);
  84                                 treeMap->readMap();
  85                                 globaldata->gTreemap = treeMap;
  86                         }
  87
  88                         namefile = validParameter.validFile(parameters, "name", true);
  89                         if (namefile == "not open") { abort = true; }
  90                         else if (namefile == "not found") { namefile = ""; }
  91                         else { readNamesFile(); }
  92
  93                         if (abort == false) {
  94                                 filename = treefile;
  95                                 read = new ReadNewickTree(filename);
  96                         }
  97
  98                 }
  99         }
 100         catch(exception& e) {
 101                 m->errorOut(e, "ReadTreeCommand", "ReadTreeCommand");
 102                 exit(1);
 103         }
 104 }
 105 //**********************************************************************************************************************
 106
 107 void ReadTreeCommand::help(){
 108         try {
 109                 m->mothurOut("The read.tree command must be run before you execute a unifrac.weighted, unifrac.unweighted. \n");
 110                 m->mothurOut("It also must be run before using the parsimony command, unless you are using the randomtree parameter.\n");
 111                 m->mothurOut("The read.tree command parameters are tree, group and name.\n");
 112                 m->mothurOut("The read.tree command should be in the following format: read.tree(tree=yourTreeFile, group=yourGroupFile).\n");
 113                 m->mothurOut("The tree and group parameters are both required.\n");
 114                 m->mothurOut("The name parameter allows you to enter a namefile.\n");
 115                 m->mothurOut("Note: No spaces between parameter labels (i.e. tree), '=' and parameters (i.e.yourTreefile).\n\n");
 116         }
 117         catch(exception& e) {
 118                 m->errorOut(e, "ReadTreeCommand", "help");
 119                 exit(1);
 120         }
 121 }
 122
 123 //**********************************************************************************************************************
 124
 125 ReadTreeCommand::~ReadTreeCommand(){
 126         if (abort == false) { delete read; }
 127 }
 128
 129 //**********************************************************************************************************************
 130
 131 int ReadTreeCommand::execute(){
 132         try {
 133
 134                 if (abort == true) { return 0; }
 135
 136                 int readOk;
 137
 138                 readOk = read->read();
 139
 140                 if (readOk != 0) { m->mothurOut("Read Terminated."); m->mothurOutEndLine(); globaldata->gTree.clear(); delete globaldata->gTreemap; return 0; }
 141
 142                 vector<Tree*> T = globaldata->gTree;
 143
 144                 //assemble users trees
 145                 for (int i = 0; i < T.size(); i++) {
 146                         if (m->control_pressed) {
 147                                 for (int i = 0; i < T.size(); i++) {  delete T[i];  }
 148                                 globaldata->gTree.clear();
 149                                 delete globaldata->gTreemap;
 150                                 return 0;
 151                         }
 152
 153                         T[i]->assembleTree();
 154                 }
 155
 156
 157 //              Sarah, isn't this checking already done when assigning the sequences to the groups?  it makes read.tree
 158 //              wicked slow...  For some reason my tree is coming in here eventhough the number of sequences in the tree
 159 //              agrees with the number of lines in the name file and the number of sequences represented by the name file
 160 //              agrees with the number of sequences in the group file
 161
 162                 //output any names that are in group file but not in tree
 163
 164 //              if (globaldata->Treenames.size() < treeMap->getNumSeqs()) {
 165 //                      cout << "in here" << endl;
 166 //                      for (int i = 0; i < treeMap->namesOfSeqs.size(); i++) {
 167 //                              //is that name in the tree?
 168 //                              int count = 0;
 169 //                              for (int j = 0; j < globaldata->Treenames.size(); j++) {
 170 //                                      if (treeMap->namesOfSeqs[i] == globaldata->Treenames[j]) { break; } //found it
 171 //                                      count++;
 172 //                              }
 173 //
 174 //                              if (m->control_pressed) {
 175 //                                      for (int i = 0; i < T.size(); i++) {  delete T[i];  }
 176 //                                      globaldata->gTree.clear();
 177 //                                      delete globaldata->gTreemap;
 178 //                                      return 0;
 179 //                              }
 180 //
 181 //                              //then you did not find it so report it
 182 //                              if (count == globaldata->Treenames.size()) {
 183 //                                      //if it is in your namefile then don't remove
 184 //                                      map<string, string>::iterator it = nameMap.find(treeMap->namesOfSeqs[i]);
 185 //
 186 //                                      if (it == nameMap.end()) {
 187 //                                              m->mothurOut(treeMap->namesOfSeqs[i] + " is in your groupfile and not in your tree. It will be disregarded."); m->mothurOutEndLine();
 188 //                                              treeMap->removeSeq(treeMap->namesOfSeqs[i]);
 189 //                                              i--; //need this because removeSeq removes name from namesOfSeqs
 190 //                                      }
 191 //                              }
 192 //                      }
 193 //
 194 //                      globaldata->gTreemap = treeMap;
 195 //              }
 196
 197                 return 0;
 198         }
 199         catch(exception& e) {
 200                 m->errorOut(e, "ReadTreeCommand", "execute");
 201                 exit(1);
 202         }
 203 }
 204 /*****************************************************************/
 205 int ReadTreeCommand::readNamesFile() {
 206         try {
 207                 globaldata->names.clear();
 208
 209                 ifstream in;
 210                 m->openInputFile(namefile, in);
 211
 212                 string first, second;
 213                 map<string, string>::iterator itNames;
 214
 215                 while(!in.eof()) {
 216                         in >> first >> second; m->gobble(in);
 217
 218                         itNames = globaldata->names.find(first);
 219                         if (itNames == globaldata->names.end()) {
 220                                 globaldata->names[first] = second;
 221
 222                                 //we need a list of names in your namefile to use above when removing extra seqs above so we don't remove them
 223                                 vector<string> dupNames;
 224                                 m->splitAtComma(second, dupNames);
 225
 226                                 for (int i = 0; i < dupNames.size(); i++) {     nameMap[dupNames[i]] = dupNames[i];  }
 227                         }else {  m->mothurOut(first + " has already been seen in namefile, disregarding names file."); m->mothurOutEndLine(); in.close(); globaldata->names.clear(); return 1; }
 228                 }
 229                 in.close();
 230
 231                 return 0;
 232         }
 233         catch(exception& e) {
 234                 m->errorOut(e, "ReadTreeCommand", "readNamesFile");
 235                 exit(1);
 236         }
 237 }
 238
 239 //**********************************************************************************************************************