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