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                 groupfile = globaldata->getGroupFile();
  19                 openInputFile(fastafile, in);
  20
  21                 if (groupfile != "") {
  22                         //read in group map info.
  23                         groupMap = new GroupMap(groupfile);
  24                         groupMap->readMap();
  25                 }
  26
  27                 fasta = new FastaMap();
  28         }
  29         catch(exception& e) {
  30                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function BinSeqCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  31                 exit(1);
  32         }
  33         catch(...) {
  34                 cout << "An unknown error has occurred in the BinSeqCommand class function BinSeqCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  35                 exit(1);
  36         }
  37 }
  38
  39 //**********************************************************************************************************************
  40
  41 BinSeqCommand::~BinSeqCommand(){
  42         delete input;
  43         delete read;
  44         delete fasta;
  45         delete list;
  46         if (groupfile != "") {
  47                 delete groupMap;
  48         }
  49 }
  50
  51 //**********************************************************************************************************************
  52
  53 int BinSeqCommand::execute(){
  54         try {
  55                 int count = 1;
  56                 string binnames, name, sequence;
  57
  58                 //read fastafile
  59                 fasta->readFastaFile(in);
  60
  61                 //set format to list so input can get listvector
  62                 globaldata->setFormat("list");
  63
  64                 //if user gave a namesfile then use it
  65                 if (namesfile != "") {
  66                         readNamesFile();
  67                 }
  68
  69                 //read list file
  70                 read = new ReadOTUFile(globaldata->getListFile());
  71                 read->read(&*globaldata);
  72
  73                 input = globaldata->ginput;
  74                 list = globaldata->gListVector;
  75
  76                 while(list != NULL){
  77
  78                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(list->getLabel()) == 1){
  79
  80                                 string outputFileName = getRootName(globaldata->getListFile()) + list->getLabel() + ".fasta";
  81                                 openOutputFile(outputFileName, out);
  82
  83                                 cout << list->getLabel() << '\t' << count << endl;
  84
  85                                 //for each bin in the list vector
  86                                 for (int i = 0; i < list->size(); i++) {
  87
  88                                         binnames = list->get(i);
  89                                         while (binnames.find_first_of(',') != -1) {
  90                                                 name = binnames.substr(0,binnames.find_first_of(','));
  91                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
  92
  93                                                 //do work for that name
  94                                                 sequence = fasta->getSequence(name);
  95                                                 if (sequence != "not found") {
  96                                                         //if you don't have groups
  97                                                         if (groupfile == "") {
  98                                                                 name = name + "|" + toString(i+1);
  99                                                                 out << ">" << name << endl;
 100                                                                 out << sequence << endl;
 101                                                         }else {//if you do have groups
 102                                                                 string group = groupMap->getGroup(name);
 103                                                                 if (group == "not found") {
 104                                                                         cout << name << " is missing from your group file. Please correct. " << endl;
 105                                                                         remove(outputFileName.c_str());
 106                                                                         return 0;
 107                                                                 }else{
 108                                                                         name = name + "|" + group + "|" + toString(i+1);
 109                                                                         out << ">" << name << endl;
 110                                                                         out << sequence << endl;
 111                                                                 }
 112                                                         }
 113                                                 }else {
 114                                                         cout << name << " is missing from your fasta or name file. Please correct. " << endl;
 115                                                         remove(outputFileName.c_str());
 116                                                         return 0;
 117                                                 }
 118
 119                                         }
 120
 121                                         //get last name
 122                                         sequence = fasta->getSequence(binnames);
 123                                         if (sequence != "not found") {
 124                                                 //if you don't have groups
 125                                                 if (groupfile == "") {
 126                                                         binnames = binnames + "|" + toString(i+1);
 127                                                         out << ">" << binnames << endl;
 128                                                         out << sequence << endl;
 129                                                 }else {//if you do have groups
 130                                                         string group = groupMap->getGroup(binnames);
 131                                                         if (group == "not found") {
 132                                                                 cout << binnames << " is missing from your group file. Please correct. " << endl;
 133                                                                 remove(outputFileName.c_str());
 134                                                                 return 0;
 135                                                         }else{
 136                                                                 binnames = binnames + "|" + group + "|" + toString(i+1);
 137                                                                 out << ">" << binnames << endl;
 138                                                                 out << sequence << endl;
 139                                                         }
 140                                                 }
 141                                         }else {
 142                                                 cout << binnames << " is missing from your fasta or name file. Please correct. " << endl;
 143                                                 remove(outputFileName.c_str());
 144                                                 return 0;
 145                                         }
 146
 147                                 }
 148                                 out.close();
 149                         }
 150
 151                         delete list;
 152                         list = input->getListVector();
 153                         count++;
 154                 }
 155
 156                 return 0;
 157         }
 158         catch(exception& e) {
 159                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 160                 exit(1);
 161         }
 162         catch(...) {
 163                 cout << "An unknown error has occurred in the BinSeqCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 164                 exit(1);
 165         }
 166 }
 167
 168 //**********************************************************************************************************************
 169 void BinSeqCommand::readNamesFile() {
 170         try {
 171                 vector<string> dupNames;
 172                 openInputFile(namesfile, inNames);
 173
 174                 string name, names, sequence;
 175
 176                 while(inNames){
 177                         inNames >> name;                        //read from first column  A
 178                         inNames >> names;               //read from second column  A,B,C,D
 179
 180                         dupNames.clear();
 181
 182                         //parse names into vector
 183                         splitAtComma(names, dupNames);
 184
 185                         //store names in fasta map
 186                         sequence = fasta->getSequence(name);
 187                         for (int i = 0; i < dupNames.size(); i++) {
 188                                 fasta->push_back(dupNames[i], sequence);
 189                         }
 190
 191                         gobble(inNames);
 192                 }
 193                 inNames.close();
 194
 195         }
 196         catch(exception& e) {
 197                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 198                 exit(1);
 199         }
 200         catch(...) {
 201                 cout << "An unknown error has occurred in the BinSeqCommand class function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 202                 exit(1);
 203         }
 204 }
 205 //**********************************************************************************************************************
 206
 207
 208