getoturepcommand.cpp

   1 /*
   2  *  getoturepcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/6/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "getoturepcommand.h"
  11
  12 //**********************************************************************************************************************
  13 GetOTURepCommand::GetOTURepCommand(){
  14         try{
  15                 globaldata = GlobalData::getInstance();
  16
  17                 if(globaldata->gSparseMatrix != NULL)   {       matrix = new SparseMatrix(*globaldata->gSparseMatrix);          }
  18
  19                 //listOfNames bin 0 = first name read in distance matrix, listOfNames bin 1 = second name read in distance matrix
  20                 if(globaldata->gListVector != NULL)             {
  21                         listOfNames = new ListVector(*globaldata->gListVector);
  22
  23                         //map names to rows in sparsematrix
  24                         for (int i = 0; i < listOfNames->size(); i++) {
  25                                 nameToIndex[listOfNames->get(i)] = i;
  26                         }
  27                 }else { cout << "error" << endl; }
  28
  29
  30                 fastafile = globaldata->getFastaFile();
  31                 namesfile = globaldata->getNameFile();
  32                 openInputFile(fastafile, in);
  33
  34                 fasta = new FastaMap();
  35
  36         }
  37         catch(exception& e) {
  38                 cout << "Standard Error: " << e.what() << " has occurred in the GetOTURepCommand class Function GetOTURepCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  39                 exit(1);
  40         }
  41         catch(...) {
  42                 cout << "An unknown error has occurred in the GetOTURepCommand class function GetOTURepCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  43                 exit(1);
  44         }
  45 }
  46 //**********************************************************************************************************************
  47
  48 GetOTURepCommand::~GetOTURepCommand(){
  49         delete matrix;
  50         delete list;
  51         delete input;
  52         delete read;
  53         delete fasta;
  54 }
  55
  56 //**********************************************************************************************************************
  57
  58 int GetOTURepCommand::execute(){
  59         try {
  60                 int count = 1;
  61                 string nameRep, name, sequence;
  62
  63                 //read fastafile
  64                 fasta->readFastaFile(in);
  65
  66                 //set format to list so input can get listvector
  67                 globaldata->setFormat("list");
  68
  69                 //if user gave a namesfile then use it
  70                 if (namesfile != "") {
  71                         readNamesFile();
  72                 }
  73
  74                 //read list file
  75                 read = new ReadPhilFile(globaldata->getListFile());
  76                 read->read(&*globaldata);
  77
  78                 input = globaldata->ginput;
  79                 list = globaldata->gListVector;
  80
  81                 while(list != NULL){
  82
  83                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(list->getLabel()) == 1){
  84
  85                                 //create output file
  86                                 string outputFileName = getRootName(globaldata->getListFile()) + list->getLabel() + ".rep.fasta";
  87                                 openOutputFile(outputFileName, out);
  88
  89                                 cout << list->getLabel() << '\t' << count << endl;
  90
  91                                 //for each bin in the list vector
  92                                 for (int i = 0; i < list->size(); i++) {
  93                                         nameRep = FindRep(i);
  94
  95                                         //print out name and sequence for that bin
  96                                         sequence = fasta->getSequence(nameRep);
  97
  98                                         if (sequence != "not found") {
  99                                                 nameRep = nameRep + "|" + toString(i+1);
 100                                                 out << ">" << nameRep << endl;
 101                                                 out << sequence << endl;
 102                                         }else {
 103                                                 cout << nameRep << " is missing from your fasta or name file. Please correct. " << endl;
 104                                                 remove(outputFileName.c_str());
 105                                                 return 0;
 106                                         }
 107                                 }
 108
 109                                 out.close();
 110                         }
 111
 112                         list = input->getListVector();
 113                         count++;
 114                 }
 115
 116
 117                 return 0;
 118         }
 119         catch(exception& e) {
 120                 cout << "Standard Error: " << e.what() << " has occurred in the GetOTURepCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 121                 exit(1);
 122         }
 123         catch(...) {
 124                 cout << "An unknown error has occurred in the GetOTURepCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 125                 exit(1);
 126         }
 127
 128 }
 129
 130 //**********************************************************************************************************************
 131 void GetOTURepCommand::readNamesFile() {
 132         try {
 133                 vector<string> dupNames;
 134                 openInputFile(namesfile, inNames);
 135
 136                 string name, names, sequence;
 137
 138                 while(inNames){
 139                         inNames >> name;                        //read from first column  A
 140                         inNames >> names;               //read from second column  A,B,C,D
 141
 142                         dupNames.clear();
 143
 144                         //parse names into vector
 145                         splitAtComma(names, dupNames);
 146
 147                         //store names in fasta map
 148                         sequence = fasta->getSequence(name);
 149                         for (int i = 0; i < dupNames.size(); i++) {
 150                                 fasta->push_back(dupNames[i], sequence);
 151                         }
 152
 153                         gobble(inNames);
 154                 }
 155                 inNames.close();
 156
 157         }
 158         catch(exception& e) {
 159                 cout << "Standard Error: " << e.what() << " has occurred in the GetOTURepCommand class Function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 160                 exit(1);
 161         }
 162         catch(...) {
 163                 cout << "An unknown error has occurred in the GetOTURepCommand class function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 164                 exit(1);
 165         }
 166 }
 167 //**********************************************************************************************************************
 168 string GetOTURepCommand::FindRep(int bin) {
 169         try{
 170                 vector<string> names;
 171                 map<string, float> sums;
 172                 map<string, float>::iterator it4;
 173                 map<int, string> binMap; //subset of namesToIndex - just member of this bin
 174                 string binnames;
 175                 float min = 10000;
 176                 string minName;
 177
 178                 binnames = list->get(bin);
 179
 180                 //parse names into vector
 181                 splitAtComma(binnames, names);
 182
 183                 //if only 1 sequence in bin then that's the rep
 184                 if (names.size() == 1) { return names[0]; }
 185                 else {
 186                         //fill binMap
 187                         for (int i = 0; i < names.size(); i++) {
 188                                 for (it3 = nameToIndex.begin(); it3 != nameToIndex.end(); it3++) {
 189                                         if (it3->first == names[i]) {
 190                                                 binMap[it3->second] = it3->first;
 191
 192                                                 //initialize sums map
 193                                                 sums[it3->first] = 0.0;
 194                                                 break;
 195                                         }
 196                                 }
 197                         }
 198
 199                         //go through each cell in the sparsematrix
 200                         for(MatData currentCell = matrix->begin(); currentCell != matrix->end(); currentCell++){
 201                                 //is this a distance between 2 members of this bin
 202                                 it = binMap.find(currentCell->row);
 203                                 it2 = binMap.find(currentCell->column);
 204
 205                                 //sum the distance of the sequences in the bin to eachother
 206                                 if ((it != binMap.end()) && (it2 != binMap.end())) {
 207                                         //this is a cell that repesents the distance between to of this bins members
 208                                         sums[it->second] += currentCell->dist;
 209                                         sums[it2->second] += currentCell->dist;
 210                                 }
 211                         }
 212
 213                         //smallest sum is the representative
 214                         for (it4 = sums.begin(); it4 != sums.end(); it4++) {
 215                                 if (it4->second < min) {
 216                                         min = it4->second;
 217                                         minName = it4->first;
 218                                 }
 219
 220                         }
 221
 222                         return minName;
 223                 }
 224
 225         }
 226         catch(exception& e) {
 227                 cout << "Standard Error: " << e.what() << " has occurred in the GetOTURepCommand class Function FindRep. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 228                 exit(1);
 229         }
 230         catch(...) {
 231                 cout << "An unknown error has occurred in the GetOTURepCommand class function FindRep. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 232                 exit(1);
 233         }
 234 }
 235
 236
 237
 238
 239