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 >> secondName >> distance;      // get the row and column names and distance
  52
  53                         if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
  54
  55                         map<string,int>::iterator itA = nameMap->find(firstName);
  56                         map<string,int>::iterator itB = nameMap->find(secondName);
  57
  58                         if(itA == nameMap->end()){  m->mothurOut("AAError: Sequence '" + firstName + "' was not found in the names file, please correct\n"); exit(1);  }
  59                         if(itB == nameMap->end()){  m->mothurOut("ABError: Sequence '" + secondName + "' was not found in the names file, please correct\n"); exit(1);  }
  60
  61                         if (distance == -1) { distance = 1000000; }
  62                         else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
  63
  64                         if(distance < cutoff && itA != itB){
  65                                 if(itA->second > itB->second){
  66                     PDistCell value(itA->second, distance);
  67
  68
  69                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  70                                                 refRow = itA->second;
  71                                                 refCol = itB->second;
  72                                                 DMatrix->addCell(itB->second, value);
  73                                         }
  74                                         else if(refRow == itA->second && refCol == itB->second){
  75                                                 lt = 0;
  76                                         }
  77                                         else{
  78                                                 DMatrix->addCell(itB->second, value);
  79                                         }
  80                                 }
  81                                 else if(itA->second < itB->second){
  82                                         PDistCell value(itB->second, distance);
  83
  84                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...
  85                                                 refRow = itA->second;
  86                                                 refCol = itB->second;
  87                                                 DMatrix->addCell(itA->second, value);
  88                                         }
  89                                         else if(refRow == itB->second && refCol == itA->second){
  90                                                 lt = 0;
  91                                         }
  92                                         else{
  93                                                 DMatrix->addCell(itA->second, value);
  94                                         }
  95                                 }
  96                                 reading->update(itA->second * nseqs);
  97                         }
  98                         m->gobble(fileHandle);
  99                 }
 100
 101                 if(lt == 0){  // oops, it was square
 102
 103                         fileHandle.close();  //let's start over
 104                         DMatrix->clear();  //let's start over
 105
 106                         m->openInputFile(distFile, fileHandle);  //let's start over
 107
 108                         while(fileHandle){
 109                                 fileHandle >> firstName >> secondName >> distance;
 110
 111                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 112
 113                                 map<string,int>::iterator itA = nameMap->find(firstName);
 114                                 map<string,int>::iterator itB = nameMap->find(secondName);
 115
 116                                 if(itA == nameMap->end()){  m->mothurOut("AAError: Sequence '" + firstName + "' was not found in the names file, please correct\n"); exit(1);  }
 117                                 if(itB == nameMap->end()){  m->mothurOut("ABError: Sequence '" + secondName + "' was not found in the names file, please correct\n"); exit(1);  }
 118
 119                                 if (distance == -1) { distance = 1000000; }
 120                                 else if (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                     PDistCell value(itA->second, distance);
 124                                         DMatrix->addCell(itB->second, value);
 125                                         reading->update(itA->second * nseqs);
 126                                 }
 127
 128                                 m->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 ReadColumnMatrix::~ReadColumnMatrix(){}
 150 /***********************************************************************/
 151