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 = openInputFile(distFile, fileHandle);
  18
  19 }
  20
  21 /***********************************************************************/
  22
  23 void ReadColumnMatrix::read(NameAssignment* nameMap){
  24         try {
  25
  26                         string firstName, secondName;
  27                         float distance;
  28                         int nseqs = nameMap->size();
  29
  30                         list = new ListVector(nameMap->getListVector());
  31
  32                         Progress* reading = new Progress("Reading matrix:     ", nseqs * nseqs);
  33
  34                         int lt = 1;
  35                         int refRow = 0; //we'll keep track of one cell - Cell(refRow,refCol) - and see if it's transpose
  36                         int refCol = 0; //shows up later - Cell(refCol,refRow).  If it does, then its a square matrix
  37
  38                         //need to see if this is a square or a triangular matrix...
  39                         while(fileHandle && lt == 1){  //let's assume it's a triangular matrix...
  40
  41                                 fileHandle >> firstName >> secondName >> distance;      // get the row and column names and distance
  42
  43                                 if(nameMap->count(firstName)==0){
  44                                         cerr << "AError: Sequence '" << firstName << "' was not found in the names file, please correct\n";
  45                                 }
  46                                 if(nameMap->count(secondName)==0){
  47                                         cerr << "AError: Sequence '" << secondName << "' was not found in the names file, please correct\n";
  48                                 }
  49
  50                                 if (distance == -1) { distance = 1000000; }
  51
  52                                 if(distance < cutoff && nameMap->get(firstName) != nameMap->get(secondName)){
  53                                         if(nameMap->get(firstName) > nameMap->get(secondName)){
  54                                                 PCell value(nameMap->get(firstName), nameMap->get(secondName), distance);
  55
  56                                                 if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  57                                                         refRow = nameMap->get(firstName);
  58                                                         refCol = nameMap->get(secondName);
  59                                                         D->addCell(value);
  60                                                 }
  61                                                 else if(refRow == nameMap->get(firstName) && refCol == nameMap->get(secondName)){
  62                                                         lt = 0;
  63                                                 }
  64                                                 else{
  65                                                         D->addCell(value);
  66                                                 }
  67                                         }
  68                                         else if(nameMap->get(firstName) < nameMap->get(secondName)){
  69                                                 PCell value(nameMap->get(secondName), nameMap->get(firstName), distance);
  70
  71                                                 if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  72                                                         refRow = nameMap->get(firstName);
  73                                                         refCol = nameMap->get(secondName);
  74                                                         D->addCell(value);
  75                                                 }
  76                                                 else if(refRow == nameMap->get(secondName) && refCol == nameMap->get(firstName)){
  77                                                         lt = 0;
  78                                                 }
  79                                                 else{
  80                                                         D->addCell(value);
  81                                                 }
  82                                         }
  83                                         reading->update(nameMap->get(firstName) * nseqs);
  84                                 }
  85                                 gobble(fileHandle);
  86                         }
  87
  88                         if(lt == 0){  // oops, it was square
  89                                 fileHandle.close();  //let's start over
  90                                 D->clear();  //let's start over
  91
  92                                 openInputFile(distFile, fileHandle);  //let's start over
  93
  94                                 while(fileHandle){
  95                                         fileHandle >> firstName >> secondName >> distance;
  96
  97                                         if(nameMap->count(firstName)==0){
  98                                                 cerr << "BError: Sequence '" << firstName << "' was not found in the names file, please correct\n";
  99                                         }
 100                                         if(nameMap->count(secondName)==0){
 101                                                 cerr << "BError: Sequence '" << secondName << "' was not found in the names file, please correct\n";
 102                                         }
 103
 104                                         if (distance == -1) { distance = 1000000; }
 105
 106                                         if(distance < cutoff && nameMap->get(firstName) > nameMap->get(secondName)){
 107                                                 PCell value(nameMap->get(firstName), nameMap->get(secondName), distance);
 108                                                 D->addCell(value);
 109                                                 reading->update(nameMap->get(firstName) * nseqs);
 110                                         }
 111
 112                                         gobble(fileHandle);
 113                                 }
 114                         }
 115                 //      else if(lt == 0){
 116                 //              while(fileHandle){
 117                 //                      fileHandle >> firstName >> secondName >> distance;
 118                 //
 119                 //                      if(nameMap->count(firstName)==0){
 120                 //                              cerr << "CError: Sequence '" << firstName << "' was not found in the names file, please correct\n";
 121                 //                      }
 122                 //                      if(nameMap->count(secondName)==0){
 123                 //                              cerr << "CError: Sequence '" << secondName << "' was not found in the names file, please correct\n";
 124                 //                      }
 125                 //                      if (distance == -1) { distance = 1000000; }
 126
 127                 //                      if(distance < cutoff && (*nameMap)[firstName].second < (*nameMap)[secondName].second){
 128                 ////                            cout << (*nameMap)[secondName] << ' ' << (*nameMap)[firstName] << ' ' << distance << endl;
 129                 //                              D->addCell(Cell((*nameMap)[secondName].second, (*nameMap)[firstName].second, distance));
 130                 //                              reading->update((*nameMap)[secondName].second * nseqs);
 131                 //                      }
 132                 //
 133                 //                      gobble(fileHandle);
 134                 //              }
 135                 //      }
 136                         reading->finish();
 137                         fileHandle.close();
 138
 139                         list->setLabel("0");
 140
 141         }
 142         catch(exception& e) {
 143                 cout << "Standard Error: " << e.what() << " has occurred in the ReadColumnMatrix class Function read. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 144                 exit(1);
 145         }
 146         catch(...) {
 147                 cout << "An unknown error has occurred in the ReadColumnMatrix class function read. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 148                 exit(1);
 149         }
 150
 151 }
 152
 153 /***********************************************************************/
 154
 155 ReadColumnMatrix::~ReadColumnMatrix(){
 156         //delete D;
 157         //delete list;
 158 }
 159
 160