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 = 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 = 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 = 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 should be in the following format: read.tree(tree=yourTreeFile, group=yourGroupFile).\n");
 112                 m->mothurOut("The tree and group parameters are both required.\n");
 113                 m->mothurOut("Note: No spaces between parameter labels (i.e. tree), '=' and parameters (i.e.yourTreefile).\n\n");
 114         }
 115         catch(exception& e) {
 116                 m->errorOut(e, "ReadTreeCommand", "help");
 117                 exit(1);
 118         }
 119 }
 120
 121 //**********************************************************************************************************************
 122
 123 ReadTreeCommand::~ReadTreeCommand(){
 124         if (abort == false) { delete read; }
 125 }
 126
 127 //**********************************************************************************************************************
 128
 129 int ReadTreeCommand::execute(){
 130         try {
 131
 132                 if (abort == true) { return 0; }
 133
 134                 int readOk;
 135
 136                 readOk = read->read();
 137
 138                 if (readOk != 0) { m->mothurOut("Read Terminated."); m->mothurOutEndLine(); globaldata->gTree.clear(); delete globaldata->gTreemap; return 0; }
 139
 140                 vector<Tree*> T = globaldata->gTree;
 141
 142                 //assemble users trees
 143                 for (int i = 0; i < T.size(); i++) {
 144                         T[i]->assembleTree();
 145                 }
 146
 147                 //output any names that are in group file but not in tree
 148                 if (globaldata->Treenames.size() < treeMap->getNumSeqs()) {
 149                         for (int i = 0; i < treeMap->namesOfSeqs.size(); i++) {
 150                                 //is that name in the tree?
 151                                 int count = 0;
 152                                 for (int j = 0; j < globaldata->Treenames.size(); j++) {
 153                                         if (treeMap->namesOfSeqs[i] == globaldata->Treenames[j]) { break; } //found it
 154                                         count++;
 155                                 }
 156
 157                                 //then you did not find it so report it
 158                                 if (count == globaldata->Treenames.size()) {
 159                                         //if it is in your namefile then don't remove
 160                                         map<string, string>::iterator it = nameMap.find(treeMap->namesOfSeqs[i]);
 161
 162                                         if (it == nameMap.end()) {
 163                                                 m->mothurOut(treeMap->namesOfSeqs[i] + " is in your groupfile and not in your tree. It will be disregarded."); m->mothurOutEndLine();
 164                                                 treeMap->removeSeq(treeMap->namesOfSeqs[i]);
 165                                                 i--; //need this because removeSeq removes name from namesOfSeqs
 166                                         }
 167                                 }
 168                         }
 169                 }
 170
 171                 return 0;
 172         }
 173         catch(exception& e) {
 174                 m->errorOut(e, "ReadTreeCommand", "execute");
 175                 exit(1);
 176         }
 177 }
 178 /*****************************************************************/
 179 int ReadTreeCommand::readNamesFile() {
 180         try {
 181                 globaldata->names.clear();
 182
 183                 ifstream in;
 184                 openInputFile(namefile, in);
 185
 186                 string first, second;
 187                 map<string, string>::iterator itNames;
 188
 189                 while(!in.eof()) {
 190                         in >> first >> second; gobble(in);
 191
 192                         itNames = globaldata->names.find(first);
 193                         if (itNames == globaldata->names.end()) {
 194                                 globaldata->names[first] = second;
 195
 196                                 //we need a list of names in your namefile to use above when removing extra seqs above so we don't remove them
 197                                 vector<string> dupNames;
 198                                 splitAtComma(second, dupNames);
 199
 200                                 for (int i = 0; i < dupNames.size(); i++) {     nameMap[dupNames[i]] = dupNames[i];  }
 201                         }else {  m->mothurOut(first + " has already been seen in namefile, disregarding names file."); m->mothurOutEndLine(); in.close(); globaldata->names.clear(); return 1; }
 202                 }
 203                 in.close();
 204
 205                 return 0;
 206         }
 207         catch(exception& e) {
 208                 m->errorOut(e, "ReadTreeCommand", "readNamesFile");
 209                 exit(1);
 210         }
 211 }
 212
 213 //**********************************************************************************************************************