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 ReadOTUFile(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                                 //make new folder for bin info
  71                                 //string foldername = "/" + getRootName(globaldata->getListFile()) + list->getLabel() + ".bins/";
  72                         //      mkdir(foldername.c_str());
  73
  74                                 string outputFileName = getRootName(globaldata->getListFile()) + list->getLabel() + ".fasta";
  75                                 openOutputFile(outputFileName, out);
  76
  77                                 cout << list->getLabel() << '\t' << count << endl;
  78
  79                                 //for each bin in the list vector
  80                                 for (int i = 0; i < list->size(); i++) {
  81
  82                                         //create output file
  83                                         //string outputFileName = foldername + getRootName(globaldata->getListFile()) + "bin" + toString(i+1) + ".fasta";
  84                                         //openOutputFile(outputFileName, out);
  85
  86                                         binnames = list->get(i);
  87                                         while (binnames.find_first_of(',') != -1) {
  88                                                 name = binnames.substr(0,binnames.find_first_of(','));
  89                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
  90
  91                                                 //do work for that name
  92                                                 sequence = fasta->getSequence(name);
  93                                                 if (sequence != "not found") {
  94                                                         name = name + "|" + toString(i+1);
  95                                                         out << ">" << name << endl;
  96                                                         out << sequence << endl;
  97                                                 }else {
  98                                                         cout << name << " is missing from your fasta or name file. Please correct. " << endl;
  99                                                         remove(outputFileName.c_str());
 100                                                         return 0;
 101                                                 }
 102
 103                                         }
 104
 105                                         //get last name
 106                                         sequence = fasta->getSequence(binnames);
 107                                         if (sequence != "not found") {
 108                                                 name = binnames + '|' + toString(i+1);
 109                                                 out << ">" << name << endl;
 110                                                 out << sequence << endl;
 111                                         }else {
 112                                                 cout << binnames << " is missing from your fasta or name file. Please correct. " << endl;
 113                                                 remove(outputFileName.c_str());
 114                                                 return 0;
 115                                         }
 116                                         //out.close();
 117                                 }
 118                                 out.close();
 119                         }
 120
 121                         list = input->getListVector();
 122                         count++;
 123                 }
 124
 125                 return 0;
 126         }
 127         catch(exception& e) {
 128                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 129                 exit(1);
 130         }
 131         catch(...) {
 132                 cout << "An unknown error has occurred in the BinSeqCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 133                 exit(1);
 134         }
 135 }
 136
 137 //**********************************************************************************************************************
 138 void BinSeqCommand::readNamesFile() {
 139         try {
 140                 vector<string> dupNames;
 141                 openInputFile(namesfile, inNames);
 142
 143                 string name, names, sequence;
 144
 145                 while(inNames){
 146                         inNames >> name;                        //read from first column  A
 147                         inNames >> names;               //read from second column  A,B,C,D
 148
 149                         dupNames.clear();
 150
 151                         //parse names into vector
 152                         splitAtComma(names, dupNames);
 153
 154                         //store names in fasta map
 155                         sequence = fasta->getSequence(name);
 156                         for (int i = 0; i < dupNames.size(); i++) {
 157                                 fasta->push_back(dupNames[i], sequence);
 158                         }
 159
 160                         gobble(inNames);
 161                 }
 162                 inNames.close();
 163
 164         }
 165         catch(exception& e) {
 166                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 167                 exit(1);
 168         }
 169         catch(...) {
 170                 cout << "An unknown error has occurred in the BinSeqCommand class function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 171                 exit(1);
 172         }
 173 }
 174 //**********************************************************************************************************************
 175
 176
 177