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