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, string o, bool s){
  15                 globaldata = GlobalData::getInstance();
  16                 m = MothurOut::getInstance();
  17         distFile = distfile;
  18                 cutoff = c;
  19                 outputDir = o;
  20                 sortWanted = s;
  21                 list = NULL;
  22 }
  23
  24 /***********************************************************************/
  25
  26 int ReadCluster::read(NameAssignment* nameMap){
  27         try {
  28
  29                 if (format == "phylip") { convertPhylip2Column(nameMap); }
  30                 else { list = new ListVector(nameMap->getListVector());  }
  31
  32                 if (m->control_pressed) { return 0; }
  33
  34                 if (sortWanted) {  OutPutFile = sortFile(distFile, outputDir);  }
  35                 else {  OutPutFile = distFile;   } //for use by clusters splitMatrix to convert a phylip matrix to column
  36
  37                 return 0;
  38
  39         }
  40         catch(exception& e) {
  41                 m->errorOut(e, "ReadCluster", "read");
  42                 exit(1);
  43         }
  44 }
  45 /***********************************************************************/
  46
  47 int ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
  48         try {
  49                 //convert phylip file to column file
  50                 map<int, string> rowToName;
  51                 map<int, string>::iterator it;
  52
  53                 ifstream in;
  54                 ofstream out;
  55                 string tempFile = distFile + ".column.temp";
  56
  57                 openInputFile(distFile, in);  gobble(in);
  58                 openOutputFile(tempFile, out);
  59
  60                 float distance;
  61                 int square, nseqs;
  62                 string name;
  63                 vector<string> matrixNames;
  64
  65                 in >> nseqs >> name;
  66                 rowToName[0] = name;
  67                 matrixNames.push_back(name);
  68
  69                 if(nameMap == NULL){
  70                         list = new ListVector(nseqs);
  71                         list->set(0, name);
  72                 }
  73                 else{
  74                         list = new ListVector(nameMap->getListVector());
  75                         if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
  76                 }
  77
  78                 char d;
  79                 while((d=in.get()) != EOF){
  80
  81                         if(isalnum(d)){
  82                                 square = 1;
  83                                 in.putback(d);
  84                                 for(int i=0;i<nseqs;i++){
  85                                         in >> distance;
  86                                 }
  87                                 break;
  88                         }
  89                         if(d == '\n'){
  90                                 square = 0;
  91                                 break;
  92                         }
  93                 }
  94
  95                 if(square == 0){
  96
  97                         for(int i=1;i<nseqs;i++){
  98                                 in >> name;
  99                                 rowToName[i] = name;
 100                                 matrixNames.push_back(name);
 101
 102                                 //there's A LOT of repeated code throughout this method...
 103                                 if(nameMap == NULL){
 104                                         list->set(i, name);
 105
 106                                         for(int j=0;j<i;j++){
 107
 108                                                 if (m->control_pressed) { in.close(); out.close(); remove(tempFile.c_str()); return 0; }
 109
 110                                                 in >> distance;
 111
 112                                                 if (distance == -1) { distance = 1000000; }
 113
 114                                                 if(distance < cutoff){
 115                                                         out << i << '\t' << j << '\t' << distance << endl;
 116                                                 }
 117                                         }
 118
 119                                 }
 120                                 else{
 121                                         if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 122
 123                                         for(int j=0;j<i;j++){
 124
 125                                                 if (m->control_pressed) { in.close(); out.close(); remove(tempFile.c_str()); return 0; }
 126
 127                                                 in >> distance;
 128
 129                                                 if (distance == -1) { distance = 1000000; }
 130
 131                                                 if(distance < cutoff){
 132                                                         out << i << '\t' << j << '\t' << distance << endl;
 133                                                 }
 134                                         }
 135                                 }
 136                         }
 137                 }
 138                 else{
 139                         for(int i=1;i<nseqs;i++){
 140                                 in >> name;
 141                                 rowToName[i] = name;
 142                                 matrixNames.push_back(name);
 143
 144                                 if(nameMap == NULL){
 145                                         list->set(i, name);
 146                                         for(int j=0;j<nseqs;j++){
 147                                                 if (m->control_pressed) { in.close(); out.close(); remove(tempFile.c_str()); return 0; }
 148
 149                                                 in >> distance;
 150
 151                                                 if (distance == -1) { distance = 1000000; }
 152
 153                                                 if(distance < cutoff && j < i){
 154                                                         out << i << '\t' << j << '\t' << distance << endl;
 155                                                 }
 156                                         }
 157                                 }
 158                                 else{
 159                                         if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 160
 161                                         for(int j=0;j<nseqs;j++){
 162                                                 if (m->control_pressed) { in.close(); out.close(); remove(tempFile.c_str()); return 0; }
 163
 164                                                 in >> distance;
 165
 166                                                 if (distance == -1) { distance = 1000000; }
 167
 168                                                 if(distance < cutoff && j < i){
 169                                                         out << i << '\t' << j << '\t' << distance << endl;
 170                                                 }
 171
 172                                         }
 173                                 }
 174                         }
 175                 }
 176
 177                 list->setLabel("0");
 178                 in.close();
 179                 out.close();
 180
 181                 if(nameMap == NULL){
 182                         nameMap = new NameAssignment();
 183                         for(int i=0;i<matrixNames.size();i++){
 184                                 nameMap->push_back(matrixNames[i]);
 185                         }
 186                         globaldata->nameMap = nameMap;
 187                 }
 188
 189
 190                 ifstream in2;
 191                 ofstream out2;
 192
 193                 string outputFile = getRootName(distFile) + "column.dist";
 194                 openInputFile(tempFile, in2);
 195                 openOutputFile(outputFile, out2);
 196
 197                 int first, second;
 198                 float dist;
 199
 200                 while (in2) {
 201                         if (m->control_pressed) { in2.close(); out2.close(); remove(tempFile.c_str()); remove(outputFile.c_str()); return 0; }
 202
 203                         in2 >> first >> second >> dist;
 204                         out2 << rowToName[first] << '\t' << rowToName[second] << '\t' << dist << endl;
 205                         gobble(in2);
 206                 }
 207                 in2.close();
 208                 out2.close();
 209
 210                 remove(tempFile.c_str());
 211                 distFile = outputFile;
 212
 213                 if (m->control_pressed) {  remove(outputFile.c_str());  }
 214
 215                 return 0;
 216         }
 217         catch(exception& e) {
 218                 m->errorOut(e, "ReadCluster", "convertPhylip2Column");
 219                 exit(1);
 220         }
 221 }
 222 /***********************************************************************/
 223
 224 ReadCluster::~ReadCluster(){}
 225 /***********************************************************************/
 226