readcolumn.cpp

   1 /*\r
   2  *  readcolumn.cpp\r
   3  *  Mothur\r
   4  *\r
   5  *  Created by Sarah Westcott on 4/21/09.\r
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.\r
   7  *\r
   8  */\r
   9 \r
  10 #include "readcolumn.h"\r
  11 #include "progress.hpp"\r
  12 \r
  13 /***********************************************************************/\r
  14 \r
  15 ReadColumnMatrix::ReadColumnMatrix(string df) : distFile(df){\r
  16         \r
  17         successOpen = openInputFile(distFile, fileHandle);\r
  18         \r
  19 }\r
  20 \r
  21 /***********************************************************************/\r
  22 \r
  23 int ReadColumnMatrix::read(NameAssignment* nameMap){\r
  24         try {           \r
  25 \r
  26                 string firstName, secondName;\r
  27                 float distance;\r
  28                 int nseqs = nameMap->size();\r
  29 \r
  30                 list = new ListVector(nameMap->getListVector());\r
  31         \r
  32                 Progress* reading = new Progress("Reading matrix:     ", nseqs * nseqs);\r
  33 \r
  34                 int lt = 1;\r
  35                 int refRow = 0; //we'll keep track of one cell - Cell(refRow,refCol) - and see if it's transpose\r
  36                 int refCol = 0; //shows up later - Cell(refCol,refRow).  If it does, then its a square matrix\r
  37 \r
  38                 //need to see if this is a square or a triangular matrix...\r
  39         \r
  40                 while(fileHandle && lt == 1){  //let's assume it's a triangular matrix...\r
  41 \r
  42                 \r
  43                         fileHandle >> firstName >> secondName >> distance;      // get the row and column names and distance\r
  44                         \r
  45                         if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }\r
  46         \r
  47                         map<string,int>::iterator itA = nameMap->find(firstName);\r
  48                         map<string,int>::iterator itB = nameMap->find(secondName);\r
  49                         \r
  50                         if(itA == nameMap->end()){\r
  51                                 cerr << "AAError: Sequence '" << firstName << "' was not found in the names file, please correct\n"; exit(1);\r
  52                         }\r
  53                         if(itB == nameMap->end()){\r
  54                                 cerr << "ABError: Sequence '" << secondName << "' was not found in the names file, please correct\n"; exit(1);\r
  55                         }\r
  56 \r
  57                         if (distance == -1) { distance = 1000000; }\r
  58                         \r
  59                         if(distance < cutoff && itA != itB){\r
  60                                 if(itA->second > itB->second){\r
  61                                         PCell value(itA->second, itB->second, distance);\r
  62                         \r
  63                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...\r
  64                                                 refRow = itA->second;\r
  65                                                 refCol = itB->second;\r
  66                                                 D->addCell(value);\r
  67                                         }\r
  68                                         else if(refRow == itA->second && refCol == itB->second){\r
  69                                                 lt = 0;\r
  70                                         }\r
  71                                         else{\r
  72                                                 D->addCell(value);\r
  73                                         }\r
  74                                 }\r
  75                                 else if(itA->second < itB->second){\r
  76                                         PCell value(itB->second, itA->second, distance);\r
  77                         \r
  78                                         if(refRow == refCol){           // in other words, if we haven't loaded refRow and refCol...\r
  79                                                 refRow = itA->second;\r
  80                                                 refCol = itB->second;\r
  81                                                 D->addCell(value);\r
  82                                         }\r
  83                                         else if(refRow == itB->second && refCol == itA->second){\r
  84                                                 lt = 0;\r
  85                                         }\r
  86                                         else{\r
  87                                                 D->addCell(value);\r
  88                                         }\r
  89                                 }\r
  90                                 reading->update(itA->second * nseqs);\r
  91                         }\r
  92                         gobble(fileHandle);\r
  93                 }\r
  94 \r
  95                 if(lt == 0){  // oops, it was square\r
  96                         fileHandle.close();  //let's start over\r
  97                         D->clear();  //let's start over\r
  98                    \r
  99                         openInputFile(distFile, fileHandle);  //let's start over\r
 100 \r
 101                         while(fileHandle){\r
 102                                 fileHandle >> firstName >> secondName >> distance;\r
 103                                 \r
 104                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }\r
 105                 \r
 106                                 map<string,int>::iterator itA = nameMap->find(firstName);\r
 107                                 map<string,int>::iterator itB = nameMap->find(secondName);\r
 108                                 \r
 109                                 if(itA == nameMap->end()){\r
 110                                         cerr << "BError: Sequence '" << firstName << "' was not found in the names file, please correct\n";\r
 111                                 }\r
 112                                 if(itB == nameMap->end()){\r
 113                                         cerr << "BError: Sequence '" << secondName << "' was not found in the names file, please correct\n";\r
 114                                 }\r
 115                                 \r
 116                                 if (distance == -1) { distance = 1000000; }\r
 117                                 \r
 118                                 if(distance < cutoff && itA->second > itB->second){\r
 119                                         PCell value(itA->second, itB->second, distance);\r
 120                                         D->addCell(value);\r
 121                                         reading->update(itA->second * nseqs);\r
 122                                 }\r
 123                 \r
 124                                 gobble(fileHandle);\r
 125                         }\r
 126                 }\r
 127                 \r
 128                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }\r
 129                 \r
 130                 reading->finish();\r
 131                 fileHandle.close();\r
 132 \r
 133                 list->setLabel("0");\r
 134                 \r
 135                 return 1;\r
 136 \r
 137         }\r
 138         catch(exception& e) {\r
 139                 m->errorOut(e, "ReadColumnMatrix", "read");\r
 140                 exit(1);\r
 141         }\r
 142 }\r
 143 \r
 144 /***********************************************************************/\r
 145 \r
 146 ReadColumnMatrix::~ReadColumnMatrix(){\r
 147         //delete D;\r
 148         //delete list;\r
 149 }\r
 150 \r
 151 \r