binsequencecommand.cpp

   1 /*
   2  *  binsequencecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/3/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amhers. All rights reserved.
   7  *
   8  */
   9
  10 #include "binsequencecommand.h"
  11
  12 //**********************************************************************************************************************
  13 BinSeqCommand::BinSeqCommand(){
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16                 fastafile = globaldata->getFastaFile();
  17                 namesfile = globaldata->getNameFile();
  18                 openInputFile(fastafile, in);
  19
  20                 fasta = new FastaMap();
  21         }
  22         catch(exception& e) {
  23                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function BinSeqCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  24                 exit(1);
  25         }
  26         catch(...) {
  27                 cout << "An unknown error has occurred in the BinSeqCommand class function BinSeqCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  28                 exit(1);
  29         }
  30 }
  31
  32 //**********************************************************************************************************************
  33
  34 BinSeqCommand::~BinSeqCommand(){
  35         delete input;
  36         delete read;
  37         delete fasta;
  38         delete list;
  39 }
  40
  41 //**********************************************************************************************************************
  42
  43 int BinSeqCommand::execute(){
  44         try {
  45                 int count = 1;
  46                 string binnames, name, sequence;
  47
  48                 //read fastafile
  49                 fasta->readFastaFile(in);
  50
  51                 //set format to list so input can get listvector
  52                 globaldata->setFormat("list");
  53
  54                 //if user gave a namesfile then use it
  55                 if (namesfile != "") {
  56                         readNamesFile();
  57                 }
  58
  59                 //read list file
  60                 read = new ReadPhilFile(globaldata->getListFile());
  61                 read->read(&*globaldata);
  62
  63                 input = globaldata->ginput;
  64                 list = globaldata->gListVector;
  65
  66                 while(list != NULL){
  67
  68                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(list->getLabel()) == 1){
  69
  70                                 //create output file
  71                                 string outputFileName = getRootName(globaldata->getListFile()) + list->getLabel() + ".fasta";
  72                                 openOutputFile(outputFileName, out);
  73
  74                                 cout << list->getLabel() << '\t' << count << endl;
  75
  76                                 //for each bin in the list vector
  77                                 for (int i = 0; i < list->size(); i++) {
  78                                         binnames = list->get(i);
  79                                         while (binnames.find_first_of(',') != -1) {
  80                                                 name = binnames.substr(0,binnames.find_first_of(','));
  81                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
  82
  83                                                 //do work for that name
  84                                                 sequence = fasta->getSequence(name);
  85                                                 if (sequence != "not found") {
  86                                                         name = name + "bin" + toString(i+1);
  87                                                         out << ">" << name << endl;
  88                                                         out << sequence << endl;
  89                                                 }else {
  90                                                         cout << name << " is missing from your fasta or name file. Please correct. " << endl;
  91                                                         remove(outputFileName.c_str());
  92                                                         return 0;
  93                                                 }
  94
  95                                         }
  96
  97                                         //get last name
  98                                         sequence = fasta->getSequence(binnames);
  99                                         if (sequence != "not found") {
 100                                                 name = binnames + "bin" + toString(i+1);
 101                                                 out << ">" << name << endl;
 102                                                 out << sequence << endl;
 103                                         }else {
 104                                                 cout << binnames << " is missing from your fasta or name file. Please correct. " << endl;
 105                                                 remove(outputFileName.c_str());
 106                                                 return 0;
 107                                         }
 108                                 }
 109                         }
 110
 111                         list = input->getListVector();
 112                         count++;
 113                 }
 114
 115                 return 0;
 116         }
 117         catch(exception& e) {
 118                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 119                 exit(1);
 120         }
 121         catch(...) {
 122                 cout << "An unknown error has occurred in the BinSeqCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 123                 exit(1);
 124         }
 125 }
 126
 127 //**********************************************************************************************************************
 128 void BinSeqCommand::readNamesFile() {
 129         try {
 130                 vector<string> dupNames;
 131                 openInputFile(namesfile, inNames);
 132
 133                 string name, names, sequence;
 134
 135                 while(inNames){
 136                         inNames >> name;                        //read from first column  A
 137                         inNames >> names;               //read from second column  A,B,C,D
 138
 139                         dupNames.clear();
 140
 141                         //parse names into vector
 142                         splitAtComma(names, dupNames);
 143
 144                         //store names in fasta map
 145                         sequence = fasta->getSequence(name);
 146                         for (int i = 0; i < dupNames.size(); i++) {
 147                                 fasta->push_back(dupNames[i], sequence);
 148                         }
 149
 150                         gobble(inNames);
 151                 }
 152                 inNames.close();
 153
 154         }
 155         catch(exception& e) {
 156                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 157                 exit(1);
 158         }
 159         catch(...) {
 160                 cout << "An unknown error has occurred in the BinSeqCommand class function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 161                 exit(1);
 162         }
 163 }
 164 //**********************************************************************************************************************
 165
 166
 167