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