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 vector<string> ReadTreeCommand::getValidParameters(){
  14         try {
  15                 string Array[] =  {"tree","group","name","outputdir","inputdir"};
  16                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  17                 return myArray;
  18         }
  19         catch(exception& e) {
  20                 m->errorOut(e, "ReadTreeCommand", "getValidParameters");
  21                 exit(1);
  22         }
  23 }
  24 //**********************************************************************************************************************
  25 vector<string> ReadTreeCommand::getRequiredParameters(){
  26         try {
  27                 string Array[] =  {"tree"};
  28                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  29                 return myArray;
  30         }
  31         catch(exception& e) {
  32                 m->errorOut(e, "ReadTreeCommand", "getRequiredParameters");
  33                 exit(1);
  34         }
  35 }
  36 //**********************************************************************************************************************
  37 vector<string> ReadTreeCommand::getRequiredFiles(){
  38         try {
  39                 vector<string> myArray;
  40                 return myArray;
  41         }
  42         catch(exception& e) {
  43                 m->errorOut(e, "ReadTreeCommand", "getRequiredFiles");
  44                 exit(1);
  45         }
  46 }
  47 //**********************************************************************************************************************
  48 ReadTreeCommand::ReadTreeCommand(string option)  {
  49         try {
  50                 globaldata = GlobalData::getInstance();
  51                 abort = false;
  52
  53                 //allow user to run help
  54                 if(option == "help") { help(); abort = true; }
  55
  56                 else {
  57                         //valid paramters for this command
  58                         string Array[] =  {"tree","group","name","outputdir","inputdir"};
  59                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  60
  61                         OptionParser parser(option);
  62                         map<string, string> parameters = parser.getParameters();
  63
  64                         ValidParameters validParameter;
  65                         map<string, string>::iterator it;
  66
  67                         //check to make sure all parameters are valid for command
  68                         for (it = parameters.begin(); it != parameters.end(); it++) {
  69                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  70                         }
  71
  72                         globaldata->newRead();
  73
  74                         //if the user changes the input directory command factory will send this info to us in the output parameter
  75                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  76                         if (inputDir == "not found"){   inputDir = "";          }
  77                         else {
  78                                 string path;
  79                                 it = parameters.find("tree");
  80                                 //user has given a template file
  81                                 if(it != parameters.end()){
  82                                         path = m->hasPath(it->second);
  83                                         //if the user has not given a path then, add inputdir. else leave path alone.
  84                                         if (path == "") {       parameters["tree"] = inputDir + it->second;             }
  85                                 }
  86
  87                                 it = parameters.find("group");
  88                                 //user has given a template file
  89                                 if(it != parameters.end()){
  90                                         path = m->hasPath(it->second);
  91                                         //if the user has not given a path then, add inputdir. else leave path alone.
  92                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
  93                                 }
  94
  95                                 it = parameters.find("name");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 101                                 }
 102
 103                         }
 104
 105
 106                         //check for required parameters
 107                         treefile = validParameter.validFile(parameters, "tree", true);
 108                         if (treefile == "not open") { abort = true; }
 109                         else if (treefile == "not found") { treefile = ""; m->mothurOut("tree is a required parameter for the read.tree command."); m->mothurOutEndLine(); abort = true;  }
 110                         else {  globaldata->setTreeFile(treefile);  globaldata->setFormat("tree");      }
 111
 112                         groupfile = validParameter.validFile(parameters, "group", true);
 113                         if (groupfile == "not open") { abort = true; }
 114                         else if (groupfile == "not found") {
 115                                 groupfile = "";
 116
 117                                 m->mothurOut("You have not provided a group file. I am assumming all sequence are from the same group."); m->mothurOutEndLine();
 118
 119                                 if (treefile != "") {  Tree* tree = new Tree(treefile); delete tree;  } //extracts names from tree to make faked out groupmap
 120
 121                                 globaldata->setGroupFile(groupfile);
 122                                 //read in group map info.
 123                                 treeMap = new TreeMap();
 124                                 for (int i = 0; i < globaldata->Treenames.size(); i++) { treeMap->addSeq(globaldata->Treenames[i], "Group1"); }
 125                                 globaldata->gTreemap = treeMap;
 126
 127                         }else {
 128                                 globaldata->setGroupFile(groupfile);
 129                                 //read in group map info.
 130                                 treeMap = new TreeMap(groupfile);
 131                                 treeMap->readMap();
 132                                 globaldata->gTreemap = treeMap;
 133                         }
 134
 135                         namefile = validParameter.validFile(parameters, "name", true);
 136                         if (namefile == "not open") { abort = true; }
 137                         else if (namefile == "not found") { namefile = ""; }
 138                         else { readNamesFile(); }
 139
 140                         if (abort == false) {
 141                                 filename = treefile;
 142                                 read = new ReadNewickTree(filename);
 143                         }
 144
 145                 }
 146         }
 147         catch(exception& e) {
 148                 m->errorOut(e, "ReadTreeCommand", "ReadTreeCommand");
 149                 exit(1);
 150         }
 151 }
 152 //**********************************************************************************************************************
 153
 154 void ReadTreeCommand::help(){
 155         try {
 156                 m->mothurOut("The read.tree command must be run before you execute a unifrac.weighted, unifrac.unweighted. \n");
 157                 m->mothurOut("It also must be run before using the parsimony command, unless you are using the randomtree parameter.\n");
 158                 m->mothurOut("The read.tree command parameters are tree, group and name.\n");
 159                 m->mothurOut("The read.tree command should be in the following format: read.tree(tree=yourTreeFile, group=yourGroupFile).\n");
 160                 m->mothurOut("The tree and group parameters are both required, if no group file is given then one group is assumed.\n");
 161                 m->mothurOut("The name parameter allows you to enter a namefile.\n");
 162                 m->mothurOut("Note: No spaces between parameter labels (i.e. tree), '=' and parameters (i.e.yourTreefile).\n\n");
 163         }
 164         catch(exception& e) {
 165                 m->errorOut(e, "ReadTreeCommand", "help");
 166                 exit(1);
 167         }
 168 }
 169
 170 //**********************************************************************************************************************
 171
 172 ReadTreeCommand::~ReadTreeCommand(){
 173         if (abort == false) { delete read; }
 174 }
 175
 176 //**********************************************************************************************************************
 177
 178 int ReadTreeCommand::execute(){
 179         try {
 180
 181                 if (abort == true) { return 0; }
 182
 183                 int readOk;
 184
 185                 readOk = read->read();
 186
 187                 if (readOk != 0) { m->mothurOut("Read Terminated."); m->mothurOutEndLine(); globaldata->gTree.clear(); delete globaldata->gTreemap; return 0; }
 188
 189                 vector<Tree*> T = globaldata->gTree;
 190
 191                 //assemble users trees
 192                 for (int i = 0; i < T.size(); i++) {
 193                         if (m->control_pressed) {
 194                                 for (int i = 0; i < T.size(); i++) {  delete T[i];  }
 195                                 globaldata->gTree.clear();
 196                                 delete globaldata->gTreemap;
 197                                 return 0;
 198                         }
 199
 200                         T[i]->assembleTree();
 201                 }
 202
 203
 204                 //if you provide a namefile we will use the numNames in the namefile as long as the number of unique match the tree names size.
 205                 int numNamesInTree;
 206                 if (namefile != "")  {
 207                         if (numUniquesInName == globaldata->Treenames.size()) {  numNamesInTree = nameMap.size();  }
 208                         else {   numNamesInTree = globaldata->Treenames.size();  }
 209                 }else {  numNamesInTree = globaldata->Treenames.size();  }
 210
 211
 212                 //output any names that are in group file but not in tree
 213                 if (numNamesInTree < treeMap->getNumSeqs()) {
 214                         for (int i = 0; i < treeMap->namesOfSeqs.size(); i++) {
 215                                 //is that name in the tree?
 216                                 int count = 0;
 217                                 for (int j = 0; j < globaldata->Treenames.size(); j++) {
 218                                         if (treeMap->namesOfSeqs[i] == globaldata->Treenames[j]) { break; } //found it
 219                                         count++;
 220                                 }
 221
 222                                 if (m->control_pressed) {
 223                                         for (int i = 0; i < T.size(); i++) {  delete T[i];  }
 224                                         globaldata->gTree.clear();
 225                                         delete globaldata->gTreemap;
 226                                         return 0;
 227                                 }
 228
 229                                 //then you did not find it so report it
 230                                 if (count == globaldata->Treenames.size()) {
 231                                         //if it is in your namefile then don't remove
 232                                         map<string, string>::iterator it = nameMap.find(treeMap->namesOfSeqs[i]);
 233
 234                                         if (it == nameMap.end()) {
 235                                                 m->mothurOut(treeMap->namesOfSeqs[i] + " is in your groupfile and not in your tree. It will be disregarded."); m->mothurOutEndLine();
 236                                                 treeMap->removeSeq(treeMap->namesOfSeqs[i]);
 237                                                 i--; //need this because removeSeq removes name from namesOfSeqs
 238                                         }
 239                                 }
 240                         }
 241
 242                         globaldata->gTreemap = treeMap;
 243                 }
 244
 245                 return 0;
 246         }
 247         catch(exception& e) {
 248                 m->errorOut(e, "ReadTreeCommand", "execute");
 249                 exit(1);
 250         }
 251 }
 252 /*****************************************************************/
 253 int ReadTreeCommand::readNamesFile() {
 254         try {
 255                 globaldata->names.clear();
 256                 numUniquesInName = 0;
 257
 258                 ifstream in;
 259                 m->openInputFile(namefile, in);
 260
 261                 string first, second;
 262                 map<string, string>::iterator itNames;
 263
 264                 while(!in.eof()) {
 265                         in >> first >> second; m->gobble(in);
 266
 267                         numUniquesInName++;
 268
 269                         itNames = globaldata->names.find(first);
 270                         if (itNames == globaldata->names.end()) {
 271                                 globaldata->names[first] = second;
 272
 273                                 //we need a list of names in your namefile to use above when removing extra seqs above so we don't remove them
 274                                 vector<string> dupNames;
 275                                 m->splitAtComma(second, dupNames);
 276
 277                                 for (int i = 0; i < dupNames.size(); i++) {     nameMap[dupNames[i]] = dupNames[i];  if ((groupfile == "") && (i != 0)) { globaldata->gTreemap->addSeq(dupNames[i], "Group1"); }  }
 278                         }else {  m->mothurOut(first + " has already been seen in namefile, disregarding names file."); m->mothurOutEndLine(); in.close(); globaldata->names.clear(); namefile = ""; return 1; }
 279                 }
 280                 in.close();
 281
 282                 return 0;
 283         }
 284         catch(exception& e) {
 285                 m->errorOut(e, "ReadTreeCommand", "readNamesFile");
 286                 exit(1);
 287         }
 288 }
 289
 290 //**********************************************************************************************************************