readcolumn.cpp

   1 /*
   2  *  readcolumn.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/21/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "readcolumn.h"
  11 #include "progress.hpp"
  12
  13 /***********************************************************************/
  14
  15 ReadColumnMatrix::ReadColumnMatrix(string df) : distFile(df){
  16
  17         successOpen = m->openInputFile(distFile, fileHandle);
  18         sim = false;
  19
  20 }
  21 /***********************************************************************/
  22
  23 ReadColumnMatrix::ReadColumnMatrix(string df, bool s) : distFile(df){
  24
  25         successOpen = m->openInputFile(distFile, fileHandle);
  26         sim = s;
  27 }
  28
  29 /***********************************************************************/
  30
  31 int ReadColumnMatrix::read(NameAssignment* nameMap){
  32         try {
  33
  34                 string firstName, secondName;
  35                 float distance;
  36                 int nseqs = nameMap->size();
  37         DMatrix->resize(nseqs);
  38                 list = new ListVector(nameMap->getListVector());
  39
  40                 Progress* reading = new Progress("Reading matrix:     ", nseqs * nseqs);
  41
  42                 int lt = 1;
  43                 int refRow = 0; //we'll keep track of one cell - Cell(refRow,refCol) - and see if it's transpose
  44                 int refCol = 0; //shows up later - Cell(refCol,refRow).  If it does, then its a square matrix
  45
  46                 //need to see if this is a square or a triangular matrix...
  47
  48                 while(fileHandle && lt == 1){  //let's assume it's a triangular matrix...
  49
  50
  51                         fileHandle >> firstName; m->gobble(fileHandle);
  52             fileHandle >> secondName; m->gobble(fileHandle);
  53             fileHandle >> distance;     // get the row and column names and distance
  54
  55             if (m->debug) { cout << firstName << '\t' << secondName << '\t' << distance << endl; }
  56
  57                         if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
  58
  59                         map<string,int>::iterator itA = nameMap->find(firstName);
  60                         map<string,int>::iterator itB = nameMap->find(secondName);
  61
  62                         if(itA == nameMap->end()){  m->mothurOut("AAError: Sequence '" + firstName + "' was not found in the names file, please correct\n"); exit(1);  }
  63                         if(itB == nameMap->end()){  m->mothurOut("ABError: Sequence '" + secondName + "' was not found in the names file, please correct\n"); exit(1);  }
  64
  65                         if (distance == -1) { distance = 1000000; }
  66                         else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
  67
  68                         if(distance < cutoff && itA != itB){
  69                                 if(itA->second > itB->second){
  70                     PDistCell value(itA->second, distance);
  71
  72
  73                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  74                                                 refRow = itA->second;
  75                                                 refCol = itB->second;
  76                                                 DMatrix->addCell(itB->second, value);
  77                                         }
  78                                         else if(refRow == itA->second && refCol == itB->second){
  79                                                 lt = 0;
  80                                         }
  81                                         else{
  82                                                 DMatrix->addCell(itB->second, value);
  83                                         }
  84                                 }
  85                                 else if(itA->second < itB->second){
  86                                         PDistCell value(itB->second, distance);
  87
  88                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  89                                                 refRow = itA->second;
  90                                                 refCol = itB->second;
  91                                                 DMatrix->addCell(itA->second, value);
  92                                         }
  93                                         else if(refRow == itB->second && refCol == itA->second){
  94                                                 lt = 0;
  95                                         }
  96                                         else{
  97                                                 DMatrix->addCell(itA->second, value);
  98                                         }
  99                                 }
 100                                 reading->update(itA->second * nseqs);
 101                         }
 102                         m->gobble(fileHandle);
 103                 }
 104
 105                 if(lt == 0){  // oops, it was square
 106
 107                         fileHandle.close();  //let's start over
 108                         DMatrix->clear();  //let's start over
 109
 110                         m->openInputFile(distFile, fileHandle);  //let's start over
 111
 112                         while(fileHandle){
 113                                 fileHandle >> firstName; m->gobble(fileHandle);
 114                 fileHandle >> secondName; m->gobble(fileHandle);
 115                 fileHandle >> distance; // get the row and column names and distance
 116
 117                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 118
 119                                 map<string,int>::iterator itA = nameMap->find(firstName);
 120                                 map<string,int>::iterator itB = nameMap->find(secondName);
 121
 122                                 if(itA == nameMap->end()){  m->mothurOut("AAError: Sequence '" + firstName + "' was not found in the names file, please correct\n"); exit(1);  }
 123                                 if(itB == nameMap->end()){  m->mothurOut("ABError: Sequence '" + secondName + "' was not found in the names file, please correct\n"); exit(1);  }
 124
 125                                 if (distance == -1) { distance = 1000000; }
 126                                 else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 127
 128                                 if(distance < cutoff && itA->second > itB->second){
 129                     PDistCell value(itA->second, distance);
 130                                         DMatrix->addCell(itB->second, value);
 131                                         reading->update(itA->second * nseqs);
 132                                 }
 133
 134                                 m->gobble(fileHandle);
 135                         }
 136                 }
 137
 138                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 139
 140                 reading->finish();
 141                 fileHandle.close();
 142
 143                 list->setLabel("0");
 144
 145                 return 1;
 146
 147         }
 148         catch(exception& e) {
 149                 m->errorOut(e, "ReadColumnMatrix", "read");
 150                 exit(1);
 151         }
 152 }
 153 /***********************************************************************/
 154
 155 int ReadColumnMatrix::read(CountTable* countTable){
 156         try {
 157
 158                 string firstName, secondName;
 159                 float distance;
 160                 int nseqs = countTable->size();
 161
 162         DMatrix->resize(nseqs);
 163                 list = new ListVector(countTable->getListVector());
 164
 165                 Progress* reading = new Progress("Reading matrix:     ", nseqs * nseqs);
 166
 167                 int lt = 1;
 168                 int refRow = 0; //we'll keep track of one cell - Cell(refRow,refCol) - and see if it's transpose
 169                 int refCol = 0; //shows up later - Cell(refCol,refRow).  If it does, then its a square matrix
 170
 171                 //need to see if this is a square or a triangular matrix...
 172
 173                 while(fileHandle && lt == 1){  //let's assume it's a triangular matrix...
 174
 175
 176                         fileHandle >> firstName; m->gobble(fileHandle);
 177             fileHandle >> secondName; m->gobble(fileHandle);
 178             fileHandle >> distance;     // get the row and column names and distance
 179
 180                         if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 181
 182                         int itA = countTable->get(firstName);
 183                         int itB = countTable->get(secondName);
 184
 185             if (m->control_pressed) { exit(1); }
 186
 187                         if (distance == -1) { distance = 1000000; }
 188                         else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 189
 190                         if(distance < cutoff && itA != itB){
 191                                 if(itA > itB){
 192                     PDistCell value(itA, distance);
 193
 194
 195                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
 196                                                 refRow = itA;
 197                                                 refCol = itB;
 198                                                 DMatrix->addCell(itB, value);
 199                                         }
 200                                         else if(refRow == itA && refCol == itB){
 201                                                 lt = 0;
 202                                         }
 203                                         else{
 204                                                 DMatrix->addCell(itB, value);
 205                                         }
 206                                 }
 207                                 else if(itA < itB){
 208                                         PDistCell value(itB, distance);
 209
 210                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
 211                                                 refRow = itA;
 212                                                 refCol = itB;
 213                                                 DMatrix->addCell(itA, value);
 214                                         }
 215                                         else if(refRow == itB && refCol == itA){
 216                                                 lt = 0;
 217                                         }
 218                                         else{
 219                                                 DMatrix->addCell(itA, value);
 220                                         }
 221                                 }
 222                                 reading->update(itA * nseqs);
 223                         }
 224                         m->gobble(fileHandle);
 225                 }
 226
 227                 if(lt == 0){  // oops, it was square
 228
 229                         fileHandle.close();  //let's start over
 230                         DMatrix->clear();  //let's start over
 231
 232                         m->openInputFile(distFile, fileHandle);  //let's start over
 233
 234                         while(fileHandle){
 235                                 fileHandle >> firstName; m->gobble(fileHandle);
 236                 fileHandle >> secondName; m->gobble(fileHandle);
 237                 fileHandle >> distance; // get the row and column names and distance
 238
 239                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 240
 241                                 int itA = countTable->get(firstName);
 242                 int itB = countTable->get(secondName);
 243
 244
 245                 if (m->control_pressed) { exit(1); }
 246
 247                                 if (distance == -1) { distance = 1000000; }
 248                                 else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 249
 250                                 if(distance < cutoff && itA > itB){
 251                     PDistCell value(itA, distance);
 252                                         DMatrix->addCell(itB, value);
 253                                         reading->update(itA * nseqs);
 254                                 }
 255
 256                                 m->gobble(fileHandle);
 257                         }
 258                 }
 259
 260                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 261
 262                 reading->finish();
 263                 fileHandle.close();
 264
 265                 list->setLabel("0");
 266
 267                 return 1;
 268
 269         }
 270         catch(exception& e) {
 271                 m->errorOut(e, "ReadColumnMatrix", "read");
 272                 exit(1);
 273         }
 274 }
 275
 276 /***********************************************************************/
 277 ReadColumnMatrix::~ReadColumnMatrix(){}
 278 /***********************************************************************/
 279