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 int 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
  43                         fileHandle >> firstName >> secondName >> distance;      // get the row and column names and distance
  44
  45                         if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
  46
  47                         map<string,int>::iterator itA = nameMap->find(firstName);
  48                         map<string,int>::iterator itB = nameMap->find(secondName);
  49
  50                         if(itA == nameMap->end()){
  51                                 cerr << "AAError: Sequence '" << firstName << "' was not found in the names file, please correct\n"; exit(1);
  52                         }
  53                         if(itB == nameMap->end()){
  54                                 cerr << "ABError: Sequence '" << secondName << "' was not found in the names file, please correct\n"; exit(1);
  55                         }
  56 //if (((itA->second == 8) && (itB->second == 1588)) || ((itA->second == 1588) && (itB->second == 8))) { cout << "found it" << endl; }
  57
  58                         if (distance == -1) { distance = 1000000; }
  59                         else if (globaldata->sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
  60
  61                         if(distance < cutoff && itA != itB){
  62                                 if(itA->second > itB->second){
  63                                         PCell value(itA->second, itB->second, distance);
  64
  65                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  66                                                 refRow = itA->second;
  67                                                 refCol = itB->second;
  68                                                 D->addCell(value);
  69                                         }
  70                                         else if(refRow == itA->second && refCol == itB->second){
  71                                                 lt = 0;
  72                                         }
  73                                         else{
  74                                                 D->addCell(value);
  75                                         }
  76                                 }
  77                                 else if(itA->second < itB->second){
  78                                         PCell value(itB->second, itA->second, distance);
  79
  80                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  81                                                 refRow = itA->second;
  82                                                 refCol = itB->second;
  83                                                 D->addCell(value);
  84                                         }
  85                                         else if(refRow == itB->second && refCol == itA->second){
  86                                                 lt = 0;
  87                                         }
  88                                         else{
  89                                                 D->addCell(value);
  90                                         }
  91                                 }
  92                                 reading->update(itA->second * nseqs);
  93                         }
  94                         gobble(fileHandle);
  95                 }
  96
  97                 if(lt == 0){  // oops, it was square
  98
  99                         fileHandle.close();  //let's start over
 100                         D->clear();  //let's start over
 101
 102                         openInputFile(distFile, fileHandle);  //let's start over
 103
 104                         while(fileHandle){
 105                                 fileHandle >> firstName >> secondName >> distance;
 106
 107                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 108
 109                                 map<string,int>::iterator itA = nameMap->find(firstName);
 110                                 map<string,int>::iterator itB = nameMap->find(secondName);
 111
 112                                 if(itA == nameMap->end()){
 113                                         cerr << "BError: Sequence '" << firstName << "' was not found in the names file, please correct\n";
 114                                 }
 115                                 if(itB == nameMap->end()){
 116                                         cerr << "BError: Sequence '" << secondName << "' was not found in the names file, please correct\n";
 117                                 }
 118
 119                                 if (distance == -1) { distance = 1000000; }
 120                                 else if (globaldata->sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 121
 122                                 if(distance < cutoff && itA->second > itB->second){
 123                                         PCell value(itA->second, itB->second, distance);
 124                                         D->addCell(value);
 125                                         reading->update(itA->second * nseqs);
 126                                 }
 127
 128                                 gobble(fileHandle);
 129                         }
 130                 }
 131
 132                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 133
 134                 reading->finish();
 135                 fileHandle.close();
 136
 137                 list->setLabel("0");
 138
 139                 return 1;
 140
 141         }
 142         catch(exception& e) {
 143                 m->errorOut(e, "ReadColumnMatrix", "read");
 144                 exit(1);
 145         }
 146 }
 147
 148 /***********************************************************************/
 149
 150 ReadColumnMatrix::~ReadColumnMatrix(){
 151         //delete D;
 152         //delete list;
 153 }
 154
 155