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
  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()){
  59                                 cerr << "AAError: Sequence '" << firstName << "' was not found in the names file, please correct\n"; exit(1);
  60                         }
  61                         if(itB == nameMap->end()){
  62                                 cerr << "ABError: Sequence '" << secondName << "' was not found in the names file, please correct\n"; exit(1);
  63                         }
  64 //if (((itA->second == 8) && (itB->second == 1588)) || ((itA->second == 1588) && (itB->second == 8))) { cout << "found it" << endl; }
  65
  66                         if (distance == -1) { distance = 1000000; }
  67                         else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
  68
  69                         if(distance < cutoff && itA != itB){
  70                                 if(itA->second > itB->second){
  71                                         PCell value(itA->second, itB->second, distance);
  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                                                 D->addCell(value);
  77                                         }
  78                                         else if(refRow == itA->second && refCol == itB->second){
  79                                                 lt = 0;
  80                                         }
  81                                         else{
  82                                                 D->addCell(value);
  83                                         }
  84                                 }
  85                                 else if(itA->second < itB->second){
  86                                         PCell value(itB->second, itA->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                                                 D->addCell(value);
  92                                         }
  93                                         else if(refRow == itB->second && refCol == itA->second){
  94                                                 lt = 0;
  95                                         }
  96                                         else{
  97                                                 D->addCell(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                         D->clear();  //let's start over
 109
 110                         m->openInputFile(distFile, fileHandle);  //let's start over
 111
 112                         while(fileHandle){
 113                                 fileHandle >> firstName >> secondName >> distance;
 114
 115                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 116
 117                                 map<string,int>::iterator itA = nameMap->find(firstName);
 118                                 map<string,int>::iterator itB = nameMap->find(secondName);
 119
 120                                 if(itA == nameMap->end()){
 121                                         cerr << "BError: Sequence '" << firstName << "' was not found in the names file, please correct\n";
 122                                 }
 123                                 if(itB == nameMap->end()){
 124                                         cerr << "BError: Sequence '" << secondName << "' was not found in the names file, please correct\n";
 125                                 }
 126
 127                                 if (distance == -1) { distance = 1000000; }
 128                                 else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 129
 130                                 if(distance < cutoff && itA->second > itB->second){
 131                                         PCell value(itA->second, itB->second, distance);
 132                                         D->addCell(value);
 133                                         reading->update(itA->second * nseqs);
 134                                 }
 135
 136                                 m->gobble(fileHandle);
 137                         }
 138                 }
 139
 140                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 141
 142                 reading->finish();
 143                 fileHandle.close();
 144
 145                 list->setLabel("0");
 146
 147                 return 1;
 148
 149         }
 150         catch(exception& e) {
 151                 m->errorOut(e, "ReadColumnMatrix", "read");
 152                 exit(1);
 153         }
 154 }
 155
 156 /***********************************************************************/
 157
 158 ReadColumnMatrix::~ReadColumnMatrix(){
 159         //delete D;
 160         //delete list;
 161 }
 162
 163