formatphylip.cpp

   1 /*
   2  *  formatphylip.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 1/13/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "formatphylip.h"
  11 #include "progress.hpp"
  12
  13 /***********************************************************************/
  14 FormatPhylipMatrix::FormatPhylipMatrix(string df) : filename(df) {
  15         m->openInputFile(filename, fileHandle);
  16 }
  17 /***********************************************************************/
  18 //not using nameMap
  19 int FormatPhylipMatrix::read(NameAssignment* nameMap){
  20         try {
  21
  22                         float distance;
  23                         int square, nseqs;
  24                         string name;
  25                         ofstream out;
  26
  27                         string numTest;
  28                         fileHandle >> numTest >> name;
  29
  30                         if (!m->isContainingOnlyDigits(numTest)) { m->mothurOut("[ERROR]: expected a number and got " + numTest + ", quitting."); m->mothurOutEndLine(); exit(1); }
  31                         else { convert(numTest, nseqs); }
  32
  33
  34                         list = new ListVector(nseqs);
  35                         list->set(0, name);
  36
  37                         char d;
  38                         while((d=fileHandle.get()) != EOF){
  39
  40                                 if(isalnum(d)){  //you are square
  41                                         square = 1;
  42                                         fileHandle.close();  //reset file
  43
  44                                         //open and get through numSeqs, code below formats rest of file
  45                                         m->openInputFile(filename, fileHandle);
  46                                         fileHandle >> nseqs; m->gobble(fileHandle);
  47
  48                                         distFile = filename + ".rowFormatted";
  49                                         m->openOutputFile(distFile, out);
  50                                         break;
  51                                 }
  52                                 if(d == '\n'){
  53                                         square = 0;
  54                                         break;
  55                                 }
  56                         }
  57
  58                         Progress* reading;
  59                         reading = new Progress("Formatting matrix:     ", nseqs * nseqs);
  60
  61                         //lower triangle, so must go to column then formatted row file
  62                         if(square == 0){
  63                                 int  index = 0;
  64
  65                                 ofstream outTemp;
  66                                 string tempFile = filename + ".temp";
  67                                 m->openOutputFile(tempFile, outTemp);
  68
  69                                 //convert to square column matrix
  70                                 for(int i=1;i<nseqs;i++){
  71
  72                                         fileHandle >> name;
  73
  74                                         list->set(i, name);
  75
  76                                         for(int j=0;j<i;j++){
  77
  78                                                 if (m->control_pressed) { outTemp.close(); remove(tempFile.c_str()); fileHandle.close();  delete reading; return 0; }
  79
  80                                                 fileHandle >> distance;
  81
  82                                                 if (distance == -1) { distance = 1000000; }
  83
  84                                                 if(distance < cutoff){
  85                                                         outTemp << i << '\t' << j << '\t' << distance << endl;
  86                                                         outTemp << j << '\t' << i << '\t' << distance << endl;
  87                                                 }
  88                                                 index++;
  89                                                 reading->update(index);
  90                                         }
  91                                 }
  92                                 outTemp.close();
  93
  94                                 //format from square column to rowFormatted
  95                                 //sort file by first column so the distances for each row are together
  96                                 string outfile = m->getRootName(tempFile) + "sorted.dist.temp";
  97
  98                                 //use the unix sort
  99                                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 100                                         string command = "sort -n " + tempFile + " -o " + outfile;
 101                                         system(command.c_str());
 102                                 #else //sort using windows sort
 103                                         string command = "sort " + tempFile + " /O " + outfile;
 104                                         system(command.c_str());
 105                                 #endif
 106
 107                                 if (m->control_pressed) { remove(tempFile.c_str()); remove(outfile.c_str());  delete reading; return 0; }
 108
 109                                 //output to new file distance for each row and save positions in file where new row begins
 110                                 ifstream in;
 111                                 m->openInputFile(outfile, in);
 112
 113                                 distFile = outfile + ".rowFormatted";
 114                                 m->openOutputFile(distFile, out);
 115
 116                                 rowPos.resize(nseqs, -1);
 117                                 int currentRow;
 118                                 int first, second;
 119                                 float dist;
 120                                 map<int, float> rowMap;
 121                                 map<int, float>::iterator itRow;
 122
 123                                 //get first currentRow
 124                                 in >> first;
 125                                 currentRow = first;
 126
 127                                 string firstString = toString(first);
 128                                 for(int k = 0; k < firstString.length(); k++)  {   in.putback(firstString[k]);  }
 129
 130                                 while(!in.eof()) {
 131                                         if (m->control_pressed) { in.close(); out.close(); remove(tempFile.c_str()); remove(distFile.c_str()); remove(outfile.c_str());  delete reading; return 0; }
 132
 133                                         in >> first >> second >> dist; m->gobble(in);
 134
 135                                         if (first != currentRow) {
 136                                                 //save position in file of each new row
 137                                                 rowPos[currentRow] = out.tellp();
 138
 139                                                 out << currentRow << '\t' << rowMap.size() << '\t';
 140
 141                                                 for (itRow = rowMap.begin(); itRow != rowMap.end(); itRow++) {
 142                                                         out << itRow->first << '\t' << itRow->second << '\t';
 143                                                 }
 144                                                 out << endl;
 145
 146                                                 currentRow = first;
 147                                                 rowMap.clear();
 148
 149                                                 //save row you just read
 150                                                 rowMap[second] = dist;
 151
 152                                                 index++;
 153                                                 reading->update(index);
 154                                         }else{
 155                                                 rowMap[second] = dist;
 156                                         }
 157                                 }
 158
 159                                 //print last Row
 160                                 //save position in file of each new row
 161                                 rowPos[currentRow] = out.tellp();
 162
 163                                 out << currentRow << '\t' << rowMap.size() << '\t';
 164
 165                                 for (itRow = rowMap.begin(); itRow != rowMap.end(); itRow++) {
 166                                         out << itRow->first << '\t' << itRow->second << '\t';
 167                                 }
 168                                 out << endl;
 169
 170                                 in.close();
 171                                 out.close();
 172
 173                                 remove(tempFile.c_str());
 174                                 remove(outfile.c_str());
 175
 176                                 if (m->control_pressed) {  remove(distFile.c_str());   delete reading; return 0; }
 177
 178                         }
 179                         else{ //square matrix convert directly to formatted row file
 180                                 int index = nseqs;
 181                                 map<int, float> rowMap;
 182                                 map<int, float>::iterator itRow;
 183                                 rowPos.resize(nseqs, -1);
 184
 185                                 for(int i=0;i<nseqs;i++){
 186                                         fileHandle >> name;
 187
 188                                         list->set(i, name);
 189
 190                                         for(int j=0;j<nseqs;j++){
 191                                                 if (m->control_pressed) {  fileHandle.close(); out.close(); remove(distFile.c_str());   delete reading; return 0; }
 192
 193                                                 fileHandle >> distance;
 194
 195                                                 if (distance == -1) { distance = 1000000; }
 196
 197                                                 if((distance < cutoff) && (j != i)){
 198                                                         rowMap[j] = distance;
 199                                                 }
 200                                                 index++;
 201                                                 reading->update(index);
 202                                         }
 203
 204                                         m->gobble(fileHandle);
 205
 206                                         //save position in file of each new row
 207                                         rowPos[i] = out.tellp();
 208
 209                                         //output row to file
 210                                         out << i << '\t' << rowMap.size() << '\t';
 211                                         for (itRow = rowMap.begin(); itRow != rowMap.end(); itRow++) {
 212                                                 out << itRow->first << '\t' << itRow->second << '\t';
 213                                         }
 214                                         out << endl;
 215
 216                                         //clear map for new row's info
 217                                         rowMap.clear();
 218                                 }
 219                         }
 220                         reading->finish();
 221                         delete reading;
 222                         fileHandle.close();
 223                         out.close();
 224
 225                         if (m->control_pressed) { remove(distFile.c_str());  return 0; }
 226
 227                         list->setLabel("0");
 228
 229                         return 1;
 230
 231
 232         }
 233         catch(exception& e) {
 234                m->errorOut(e, "FormatPhylipMatrix", "read");
 235                 exit(1);
 236         }
 237 }
 238 /***********************************************************************/
 239 FormatPhylipMatrix::~FormatPhylipMatrix(){}
 240 /***********************************************************************/
 241
 242