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") {
  80                                 groupfile = "";
  81
  82                                 m->mothurOut("You have not provided a group file. I am assumming all sequence are from the same group."); m->mothurOutEndLine();
  83
  84                                 if (treefile != "") {  Tree* tree = new Tree(treefile); delete tree;  } //extracts names from tree to make faked out groupmap
  85
  86                                 globaldata->setGroupFile(groupfile);
  87                                 //read in group map info.
  88                                 treeMap = new TreeMap();
  89                                 for (int i = 0; i < globaldata->Treenames.size(); i++) { treeMap->addSeq(globaldata->Treenames[i], "Group1"); }
  90                                 globaldata->gTreemap = treeMap;
  91
  92                         }else {
  93                                 globaldata->setGroupFile(groupfile);
  94                                 //read in group map info.
  95                                 treeMap = new TreeMap(groupfile);
  96                                 treeMap->readMap();
  97                                 globaldata->gTreemap = treeMap;
  98                         }
  99
 100                         namefile = validParameter.validFile(parameters, "name", true);
 101                         if (namefile == "not open") { abort = true; }
 102                         else if (namefile == "not found") { namefile = ""; }
 103                         else { readNamesFile(); }
 104
 105                         if (abort == false) {
 106                                 filename = treefile;
 107                                 read = new ReadNewickTree(filename);
 108                         }
 109
 110                 }
 111         }
 112         catch(exception& e) {
 113                 m->errorOut(e, "ReadTreeCommand", "ReadTreeCommand");
 114                 exit(1);
 115         }
 116 }
 117 //**********************************************************************************************************************
 118
 119 void ReadTreeCommand::help(){
 120         try {
 121                 m->mothurOut("The read.tree command must be run before you execute a unifrac.weighted, unifrac.unweighted. \n");
 122                 m->mothurOut("It also must be run before using the parsimony command, unless you are using the randomtree parameter.\n");
 123                 m->mothurOut("The read.tree command parameters are tree, group and name.\n");
 124                 m->mothurOut("The read.tree command should be in the following format: read.tree(tree=yourTreeFile, group=yourGroupFile).\n");
 125                 m->mothurOut("The tree and group parameters are both required.\n");
 126                 m->mothurOut("The name parameter allows you to enter a namefile.\n");
 127                 m->mothurOut("Note: No spaces between parameter labels (i.e. tree), '=' and parameters (i.e.yourTreefile).\n\n");
 128         }
 129         catch(exception& e) {
 130                 m->errorOut(e, "ReadTreeCommand", "help");
 131                 exit(1);
 132         }
 133 }
 134
 135 //**********************************************************************************************************************
 136
 137 ReadTreeCommand::~ReadTreeCommand(){
 138         if (abort == false) { delete read; }
 139 }
 140
 141 //**********************************************************************************************************************
 142
 143 int ReadTreeCommand::execute(){
 144         try {
 145
 146                 if (abort == true) { return 0; }
 147
 148                 int readOk;
 149
 150                 readOk = read->read();
 151
 152                 if (readOk != 0) { m->mothurOut("Read Terminated."); m->mothurOutEndLine(); globaldata->gTree.clear(); delete globaldata->gTreemap; return 0; }
 153
 154                 vector<Tree*> T = globaldata->gTree;
 155
 156                 //assemble users trees
 157                 for (int i = 0; i < T.size(); i++) {
 158                         if (m->control_pressed) {
 159                                 for (int i = 0; i < T.size(); i++) {  delete T[i];  }
 160                                 globaldata->gTree.clear();
 161                                 delete globaldata->gTreemap;
 162                                 return 0;
 163                         }
 164
 165                         T[i]->assembleTree();
 166                 }
 167
 168
 169                 //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.
 170                 int numNamesInTree;
 171                 if (namefile != "")  {
 172                         if (numUniquesInName == globaldata->Treenames.size()) {  numNamesInTree = nameMap.size();  }
 173                         else {   numNamesInTree = globaldata->Treenames.size();  }
 174                 }else {  numNamesInTree = globaldata->Treenames.size();  }
 175
 176
 177                 //output any names that are in group file but not in tree
 178                 if (numNamesInTree < treeMap->getNumSeqs()) {
 179                         for (int i = 0; i < treeMap->namesOfSeqs.size(); i++) {
 180                                 //is that name in the tree?
 181                                 int count = 0;
 182                                 for (int j = 0; j < globaldata->Treenames.size(); j++) {
 183                                         if (treeMap->namesOfSeqs[i] == globaldata->Treenames[j]) { break; } //found it
 184                                         count++;
 185                                 }
 186
 187                                 if (m->control_pressed) {
 188                                         for (int i = 0; i < T.size(); i++) {  delete T[i];  }
 189                                         globaldata->gTree.clear();
 190                                         delete globaldata->gTreemap;
 191                                         return 0;
 192                                 }
 193
 194                                 //then you did not find it so report it
 195                                 if (count == globaldata->Treenames.size()) {
 196                                         //if it is in your namefile then don't remove
 197                                         map<string, string>::iterator it = nameMap.find(treeMap->namesOfSeqs[i]);
 198
 199                                         if (it == nameMap.end()) {
 200                                                 m->mothurOut(treeMap->namesOfSeqs[i] + " is in your groupfile and not in your tree. It will be disregarded."); m->mothurOutEndLine();
 201                                                 treeMap->removeSeq(treeMap->namesOfSeqs[i]);
 202                                                 i--; //need this because removeSeq removes name from namesOfSeqs
 203                                         }
 204                                 }
 205                         }
 206
 207                         globaldata->gTreemap = treeMap;
 208                 }
 209
 210                 return 0;
 211         }
 212         catch(exception& e) {
 213                 m->errorOut(e, "ReadTreeCommand", "execute");
 214                 exit(1);
 215         }
 216 }
 217 /*****************************************************************/
 218 int ReadTreeCommand::readNamesFile() {
 219         try {
 220                 globaldata->names.clear();
 221                 numUniquesInName = 0;
 222
 223                 ifstream in;
 224                 m->openInputFile(namefile, in);
 225
 226                 string first, second;
 227                 map<string, string>::iterator itNames;
 228
 229                 while(!in.eof()) {
 230                         in >> first >> second; m->gobble(in);
 231
 232                         numUniquesInName++;
 233
 234                         itNames = globaldata->names.find(first);
 235                         if (itNames == globaldata->names.end()) {
 236                                 globaldata->names[first] = second;
 237
 238                                 //we need a list of names in your namefile to use above when removing extra seqs above so we don't remove them
 239                                 vector<string> dupNames;
 240                                 m->splitAtComma(second, dupNames);
 241
 242                                 for (int i = 0; i < dupNames.size(); i++) {     nameMap[dupNames[i]] = dupNames[i];  if ((groupfile == "") && (i != 0)) { globaldata->gTreemap->addSeq(dupNames[i], "Group1"); }  }
 243                         }else {  m->mothurOut(first + " has already been seen in namefile, disregarding names file."); m->mothurOutEndLine(); in.close(); globaldata->names.clear(); namefile = ""; return 1; }
 244                 }
 245                 in.close();
 246
 247                 return 0;
 248         }
 249         catch(exception& e) {
 250                 m->errorOut(e, "ReadTreeCommand", "readNamesFile");
 251                 exit(1);
 252         }
 253 }
 254
 255 //**********************************************************************************************************************