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
  40                 while(fileHandle && lt == 1){  //let's assume it's a triangular matrix...
  41
  42                         fileHandle >> firstName >> secondName >> distance;      // get the row and column names and distance
  43
  44                         map<string,int>::iterator itA = nameMap->find(firstName);
  45                         map<string,int>::iterator itB = nameMap->find(secondName);
  46
  47                         if(itA == nameMap->end()){
  48                                 cerr << "AAError: Sequence '" << firstName << "' was not found in the names file, please correct\n";
  49                         }
  50                         if(itB == nameMap->end()){
  51                                 cerr << "ABError: Sequence '" << secondName << "' was not found in the names file, please correct\n";
  52                         }
  53
  54                         if (distance == -1) { distance = 1000000; }
  55
  56                         if(distance < cutoff && itA != itB){
  57                                 if(itA->second > itB->second){
  58                                         PCell value(itA->second, itB->second, distance);
  59
  60                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  61                                                 refRow = itA->second;
  62                                                 refCol = itB->second;
  63                                                 D->addCell(value);
  64                                         }
  65                                         else if(refRow == itA->second && refCol == itB->second){
  66                                                 lt = 0;
  67                                         }
  68                                         else{
  69                                                 D->addCell(value);
  70                                         }
  71                                 }
  72                                 else if(itA->second < itB->second){
  73                                         PCell value(itB->second, itA->second, distance);
  74
  75                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  76                                                 refRow = itA->second;
  77                                                 refCol = itB->second;
  78                                                 D->addCell(value);
  79                                         }
  80                                         else if(refRow == itB->second && refCol == itA->second){
  81                                                 lt = 0;
  82                                         }
  83                                         else{
  84                                                 D->addCell(value);
  85                                         }
  86                                 }
  87                                 reading->update(itA->second * nseqs);
  88                         }
  89                         gobble(fileHandle);
  90                 }
  91
  92                 if(lt == 0){  // oops, it was square
  93                         fileHandle.close();  //let's start over
  94                         D->clear();  //let's start over
  95
  96                         openInputFile(distFile, fileHandle);  //let's start over
  97
  98                         while(fileHandle){
  99                                 fileHandle >> firstName >> secondName >> distance;
 100
 101                                 map<string,int>::iterator itA = nameMap->find(firstName);
 102                                 map<string,int>::iterator itB = nameMap->find(secondName);
 103
 104                                 if(itA == nameMap->end()){
 105                                         cerr << "BError: Sequence '" << firstName << "' was not found in the names file, please correct\n";
 106                                 }
 107                                 if(itB == nameMap->end()){
 108                                         cerr << "BError: Sequence '" << secondName << "' was not found in the names file, please correct\n";
 109                                 }
 110
 111                                 if (distance == -1) { distance = 1000000; }
 112
 113                                 if(distance < cutoff && itA->second > itB->second){
 114                                         PCell value(itA->second, itB->second, distance);
 115                                         D->addCell(value);
 116                                         reading->update(itA->second * nseqs);
 117                                 }
 118
 119                                 gobble(fileHandle);
 120                         }
 121                 }
 122
 123                 reading->finish();
 124                 fileHandle.close();
 125
 126                 list->setLabel("0");
 127
 128         }
 129         catch(exception& e) {
 130                 cout << "Standard Error: " << e.what() << " has occurred in the ReadColumnMatrix class Function read. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 131                 exit(1);
 132         }
 133         catch(...) {
 134                 cout << "An unknown error has occurred in the ReadColumnMatrix class function read. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 135                 exit(1);
 136         }
 137
 138 }
 139
 140 /***********************************************************************/
 141
 142 ReadColumnMatrix::~ReadColumnMatrix(){
 143         //delete D;
 144         //delete list;
 145 }
 146
 147