readphylip.cpp

   1 /*
   2  *  readphylip.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 "readphylip.h"
  11 #include "progress.hpp"
  12
  13 /***********************************************************************/
  14
  15 ReadPhylipMatrix::ReadPhylipMatrix(string distFile){
  16
  17         successOpen = openInputFile(distFile, fileHandle);
  18
  19 }
  20
  21 /***********************************************************************/
  22
  23 void ReadPhylipMatrix::read(NameAssignment* nameMap){
  24         try {
  25
  26                         float distance;
  27                         int square, nseqs;
  28                         string name;
  29                         vector<string> matrixNames;
  30
  31                         fileHandle >> nseqs >> name;
  32
  33                         matrixNames.push_back(name);
  34
  35                         if(nameMap == NULL){
  36                                 list = new ListVector(nseqs);
  37                                 list->set(0, name);
  38                         }
  39                         else{
  40                                 list = new ListVector(nameMap->getListVector());
  41                                 if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
  42                         }
  43
  44                         char d;
  45                         while((d=fileHandle.get()) != EOF){
  46
  47                                 if(isalnum(d)){
  48                                         square = 1;
  49                                         fileHandle.putback(d);
  50                                         for(int i=0;i<nseqs;i++){
  51                                                 fileHandle >> distance;
  52                                         }
  53                                         break;
  54                                 }
  55                                 if(d == '\n'){
  56                                         square = 0;
  57                                         break;
  58                                 }
  59                         }
  60
  61                         Progress* reading;
  62
  63                         if(square == 0){
  64
  65                                 reading = new Progress("Reading matrix:     ", nseqs * (nseqs - 1) / 2);
  66
  67                                 int        index = 0;
  68
  69                                 for(int i=1;i<nseqs;i++){
  70                                         fileHandle >> name;
  71                                         matrixNames.push_back(name);
  72
  73                                         //there's A LOT of repeated code throughout this method...
  74                                         if(nameMap == NULL){
  75                                                 list->set(i, name);
  76
  77                                                 for(int j=0;j<i;j++){
  78                                                         fileHandle >> distance;
  79
  80                                                         if (distance == -1) { distance = 1000000; }
  81
  82                                                         if(distance < cutoff){
  83                                                                 PCell value(i, j, distance);
  84                                                                 D->addCell(value);
  85                                                         }
  86                                                         index++;
  87                                                         reading->update(index);
  88                                                 }
  89
  90                                         }
  91                                         else{
  92                                                 if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
  93
  94                                                 for(int j=0;j<i;j++){
  95                                                         fileHandle >> distance;
  96
  97                                                         if (distance == -1) { distance = 1000000; }
  98
  99                                                         if(distance < cutoff){
 100                                                                 PCell value(nameMap->get(matrixNames[i]), nameMap->get(matrixNames[j]), distance);
 101                                                                 D->addCell(value);
 102                                                         }
 103                                                         index++;
 104                                                         reading->update(index);
 105                                                 }
 106                                         }
 107                                 }
 108                         }
 109                         else{
 110
 111                                 reading = new Progress("Reading matrix:     ", nseqs * nseqs);
 112
 113                                 int index = nseqs;
 114
 115                                 for(int i=1;i<nseqs;i++){
 116                                         fileHandle >> name;
 117                                         matrixNames.push_back(name);
 118
 119                                         if(nameMap == NULL){
 120                                                 list->set(i, name);
 121                                                 for(int j=0;j<nseqs;j++){
 122                                                         fileHandle >> distance;
 123
 124                                                         if (distance == -1) { distance = 1000000; }
 125
 126                                                         if(distance < cutoff && j < i){
 127                                                                 PCell value(i, j, distance);
 128                                                                 D->addCell(value);
 129                                                         }
 130                                                         index++;
 131                                                         reading->update(index);
 132                                                 }
 133
 134                                         }
 135                                         else{
 136                                                 if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 137
 138                                                 for(int j=0;j<nseqs;j++){
 139                                                         fileHandle >> distance;
 140
 141                                                         if (distance == -1) { distance = 1000000; }
 142
 143                                                         if(distance < cutoff && j < i){
 144                                                                 PCell value(nameMap->get(matrixNames[i]), nameMap->get(matrixNames[j]), distance);
 145                                                                 D->addCell(value);
 146                                                         }
 147                                                         index++;
 148                                                         reading->update(index);
 149                                                 }
 150                                         }
 151                                 }
 152                         }
 153                         reading->finish();
 154                         delete reading;
 155
 156                         list->setLabel("0");
 157                         fileHandle.close();
 158
 159                      /*   if(nameMap != NULL){
 160                                 for(int i=0;i<matrixNames.size();i++){
 161                                         nameMap->erase(matrixNames[i]);
 162                                 }
 163                                 if(nameMap->size() > 0){
 164                                         //should probably tell them what is missing if we missed something
 165                                         m->mothurOut("missed something\t" + toString(nameMap->size())); m->mothurOutEndLine();
 166                                 }
 167                         } */
 168
 169                 }
 170         catch(exception& e) {
 171                m->errorOut(e, "ReadPhylipMatrix", "read");
 172                 exit(1);
 173         }
 174         }
 175
 176 /***********************************************************************/
 177
 178 ReadPhylipMatrix::~ReadPhylipMatrix(){
 179        // delete D;
 180        // delete list;
 181 }