readcluster.cpp

   1 /*
   2  *  readcluster.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/28/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "readcluster.h"
  11
  12 /***********************************************************************/
  13
  14 ReadCluster::ReadCluster(string distfile, float c){
  15                 globaldata = GlobalData::getInstance();
  16                 m = MothurOut::getInstance();
  17         distFile = distfile;
  18                 cutoff = c;
  19 }
  20
  21 /***********************************************************************/
  22
  23 int ReadCluster::read(NameAssignment* nameMap){
  24         try {
  25
  26                 if (format == "phylip") { convertPhylip2Column(nameMap); }
  27                 else { list = new ListVector(nameMap->getListVector());  }
  28
  29                 if (m->control_pressed) { return 0; }
  30
  31                 OutPutFile = sortFile(distFile);
  32
  33                 return 0;
  34
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "ReadCluster", "read");
  38                 exit(1);
  39         }
  40 }
  41 /***********************************************************************/
  42
  43 int ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
  44         try {
  45                 //convert phylip file to column file
  46                 map<int, string> rowToName;
  47                 map<int, string>::iterator it;
  48
  49                 ifstream in;
  50                 ofstream out;
  51                 string tempFile = distFile + ".column.temp";
  52
  53                 openInputFile(distFile, in);
  54                 openOutputFile(tempFile, out);
  55
  56                 float distance;
  57                 int square, nseqs;
  58                 string name;
  59                 vector<string> matrixNames;
  60
  61                 in >> nseqs >> name;
  62                 rowToName[0] = name;
  63                 matrixNames.push_back(name);
  64
  65                 if(nameMap == NULL){
  66                         list = new ListVector(nseqs);
  67                         list->set(0, name);
  68                 }
  69                 else{
  70                         list = new ListVector(nameMap->getListVector());
  71                         if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
  72                 }
  73
  74                 char d;
  75                 while((d=in.get()) != EOF){
  76
  77                         if(isalnum(d)){
  78                                 square = 1;
  79                                 in.putback(d);
  80                                 for(int i=0;i<nseqs;i++){
  81                                         in >> distance;
  82                                 }
  83                                 break;
  84                         }
  85                         if(d == '\n'){
  86                                 square = 0;
  87                                 break;
  88                         }
  89                 }
  90
  91                 if(square == 0){
  92
  93                         for(int i=1;i<nseqs;i++){
  94                                 in >> name;
  95                                 rowToName[i] = name;
  96                                 matrixNames.push_back(name);
  97
  98                                 //there's A LOT of repeated code throughout this method...
  99                                 if(nameMap == NULL){
 100                                         list->set(i, name);
 101
 102                                         for(int j=0;j<i;j++){
 103
 104                                                 if (m->control_pressed) { in.close(); out.close(); remove(tempFile.c_str()); return 0; }
 105
 106                                                 in >> distance;
 107
 108                                                 if (distance == -1) { distance = 1000000; }
 109
 110                                                 if(distance < cutoff){
 111                                                         out << i << '\t' << j << '\t' << distance << endl;
 112                                                 }
 113                                         }
 114
 115                                 }
 116                                 else{
 117                                         if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 118
 119                                         for(int j=0;j<i;j++){
 120
 121                                                 if (m->control_pressed) { in.close(); out.close(); remove(tempFile.c_str()); return 0; }
 122
 123                                                 in >> distance;
 124
 125                                                 if (distance == -1) { distance = 1000000; }
 126
 127                                                 if(distance < cutoff){
 128                                                         out << i << '\t' << j << '\t' << distance << endl;
 129                                                 }
 130                                         }
 131                                 }
 132                         }
 133                 }
 134                 else{
 135                         for(int i=1;i<nseqs;i++){
 136                                 in >> name;
 137                                 rowToName[i] = name;
 138                                 matrixNames.push_back(name);
 139
 140                                 if(nameMap == NULL){
 141                                         list->set(i, name);
 142                                         for(int j=0;j<nseqs;j++){
 143                                                 if (m->control_pressed) { in.close(); out.close(); remove(tempFile.c_str()); return 0; }
 144
 145                                                 in >> distance;
 146
 147                                                 if (distance == -1) { distance = 1000000; }
 148
 149                                                 if(distance < cutoff && j < i){
 150                                                         out << i << '\t' << j << '\t' << distance << endl;
 151                                                 }
 152                                         }
 153                                 }
 154                                 else{
 155                                         if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 156
 157                                         for(int j=0;j<nseqs;j++){
 158                                                 if (m->control_pressed) { in.close(); out.close(); remove(tempFile.c_str()); return 0; }
 159
 160                                                 in >> distance;
 161
 162                                                 if (distance == -1) { distance = 1000000; }
 163
 164                                                 if(distance < cutoff && j < i){
 165                                                         out << i << '\t' << j << '\t' << distance << endl;
 166                                                 }
 167
 168                                         }
 169                                 }
 170                         }
 171                 }
 172
 173                 list->setLabel("0");
 174                 in.close();
 175                 out.close();
 176
 177                 if(nameMap == NULL){
 178                         nameMap = new NameAssignment();
 179                         for(int i=0;i<matrixNames.size();i++){
 180                                 nameMap->push_back(matrixNames[i]);
 181                         }
 182                         globaldata->nameMap = nameMap;
 183                 }
 184
 185
 186                 ifstream in2;
 187                 ofstream out2;
 188
 189                 string outputFile = getRootName(distFile) + "column.dist";
 190                 openInputFile(tempFile, in2);
 191                 openOutputFile(outputFile, out2);
 192
 193                 int first, second;
 194                 float dist;
 195
 196                 while (in2) {
 197                         if (m->control_pressed) { in2.close(); out2.close(); remove(tempFile.c_str()); remove(outputFile.c_str()); return 0; }
 198
 199                         in2 >> first >> second >> dist;
 200                         out2 << rowToName[first] << '\t' << rowToName[second] << '\t' << dist << endl;
 201                         gobble(in2);
 202                 }
 203                 in2.close();
 204                 out2.close();
 205
 206                 remove(tempFile.c_str());
 207                 distFile = outputFile;
 208
 209                 if (m->control_pressed) {  remove(outputFile.c_str());  }
 210
 211                 return 0;
 212         }
 213         catch(exception& e) {
 214                 m->errorOut(e, "ReadCluster", "convertPhylip2Column");
 215                 exit(1);
 216         }
 217 }
 218 /***********************************************************************/
 219
 220 ReadCluster::~ReadCluster(){}
 221 /***********************************************************************/
 222