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                 int error = 0;
  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                 ListVector* lastList = list;
  76
  77                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
  78                 set<string> processedLabels;
  79                 set<string> userLabels = globaldata->labels;
  80
  81
  82                 while((list != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0))) {
  83
  84                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(list->getLabel()) == 1){
  85
  86                                 error = process(list, count);
  87                                 if (error == 1) { return 0; }
  88
  89                                 processedLabels.insert(list->getLabel());
  90                                 userLabels.erase(list->getLabel());
  91
  92                         }
  93
  94                         if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastList->getLabel()) != 1)) {
  95
  96                                 error = process(lastList, count);
  97                                 if (error == 1) { return 0; }
  98
  99                                 processedLabels.insert(lastList->getLabel());
 100                                 userLabels.erase(lastList->getLabel());
 101
 102                         }
 103
 104                         if (count != 1) { delete lastList; }
 105                         lastList = list;
 106
 107                         list = input->getListVector();
 108                         count++;
 109                 }
 110
 111
 112                 //output error messages about any remaining user labels
 113                 set<string>::iterator it;
 114                 bool needToRun = false;
 115                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 116                         cout << "Your file does not include the label "<< *it;
 117                         if (processedLabels.count(lastList->getLabel()) != 1) {
 118                                 cout << ". I will use " << lastList->getLabel() << "." << endl;
 119                                 needToRun = true;
 120                         }else {
 121                                 cout << ". Please refer to " << lastList->getLabel() << "." << endl;
 122                         }
 123                 }
 124
 125                 //run last line if you need to
 126                 if (needToRun == true)  {
 127                         error = process(lastList, count);
 128                         if (error == 1) { return 0; }
 129                 }
 130
 131                 delete lastList;
 132                 return 0;
 133         }
 134         catch(exception& e) {
 135                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 136                 exit(1);
 137         }
 138         catch(...) {
 139                 cout << "An unknown error has occurred in the BinSeqCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 140                 exit(1);
 141         }
 142 }
 143
 144 //**********************************************************************************************************************
 145 void BinSeqCommand::readNamesFile() {
 146         try {
 147                 vector<string> dupNames;
 148                 openInputFile(namesfile, inNames);
 149
 150                 string name, names, sequence;
 151
 152                 while(inNames){
 153                         inNames >> name;                        //read from first column  A
 154                         inNames >> names;               //read from second column  A,B,C,D
 155
 156                         dupNames.clear();
 157
 158                         //parse names into vector
 159                         splitAtComma(names, dupNames);
 160
 161                         //store names in fasta map
 162                         sequence = fasta->getSequence(name);
 163                         for (int i = 0; i < dupNames.size(); i++) {
 164                                 fasta->push_back(dupNames[i], sequence);
 165                         }
 166
 167                         gobble(inNames);
 168                 }
 169                 inNames.close();
 170
 171         }
 172         catch(exception& e) {
 173                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 174                 exit(1);
 175         }
 176         catch(...) {
 177                 cout << "An unknown error has occurred in the BinSeqCommand class function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 178                 exit(1);
 179         }
 180 }
 181 //**********************************************************************************************************************
 182 //return 1 if error, 0 otherwise
 183 int BinSeqCommand::process(ListVector* list, int count) {
 184         try {
 185                                 string binnames, name, sequence;
 186                                 string outputFileName = getRootName(globaldata->getListFile()) + list->getLabel() + ".fasta";
 187                                 openOutputFile(outputFileName, out);
 188
 189                                 cout << list->getLabel() << '\t' << count << endl;
 190
 191                                 //for each bin in the list vector
 192                                 for (int i = 0; i < list->size(); i++) {
 193
 194                                         binnames = list->get(i);
 195                                         while (binnames.find_first_of(',') != -1) {
 196                                                 name = binnames.substr(0,binnames.find_first_of(','));
 197                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 198
 199                                                 //do work for that name
 200                                                 sequence = fasta->getSequence(name);
 201                                                 if (sequence != "not found") {
 202                                                         //if you don't have groups
 203                                                         if (groupfile == "") {
 204                                                                 name = name + "|" + toString(i+1);
 205                                                                 out << ">" << name << endl;
 206                                                                 out << sequence << endl;
 207                                                         }else {//if you do have groups
 208                                                                 string group = groupMap->getGroup(name);
 209                                                                 if (group == "not found") {
 210                                                                         cout << name << " is missing from your group file. Please correct. " << endl;
 211                                                                         remove(outputFileName.c_str());
 212                                                                         return 1;
 213                                                                 }else{
 214                                                                         name = name + "|" + group + "|" + toString(i+1);
 215                                                                         out << ">" << name << endl;
 216                                                                         out << sequence << endl;
 217                                                                 }
 218                                                         }
 219                                                 }else {
 220                                                         cout << name << " is missing from your fasta or name file. Please correct. " << endl;
 221                                                         remove(outputFileName.c_str());
 222                                                         return 1;
 223                                                 }
 224
 225                                         }
 226
 227                                         //get last name
 228                                         sequence = fasta->getSequence(binnames);
 229                                         if (sequence != "not found") {
 230                                                 //if you don't have groups
 231                                                 if (groupfile == "") {
 232                                                         binnames = binnames + "|" + toString(i+1);
 233                                                         out << ">" << binnames << endl;
 234                                                         out << sequence << endl;
 235                                                 }else {//if you do have groups
 236                                                         string group = groupMap->getGroup(binnames);
 237                                                         if (group == "not found") {
 238                                                                 cout << binnames << " is missing from your group file. Please correct. " << endl;
 239                                                                 remove(outputFileName.c_str());
 240                                                                 return 1;
 241                                                         }else{
 242                                                                 binnames = binnames + "|" + group + "|" + toString(i+1);
 243                                                                 out << ">" << binnames << endl;
 244                                                                 out << sequence << endl;
 245                                                         }
 246                                                 }
 247                                         }else {
 248                                                 cout << binnames << " is missing from your fasta or name file. Please correct. " << endl;
 249                                                 remove(outputFileName.c_str());
 250                                                 return 1;
 251                                         }
 252                                 }
 253
 254                                 out.close();
 255                                 return 0;
 256
 257         }
 258         catch(exception& e) {
 259                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 260                 exit(1);
 261         }
 262         catch(...) {
 263                 cout << "An unknown error has occurred in the BinSeqCommand class function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 264                 exit(1);
 265         }
 266 }
 267 //**********************************************************************************************************************
 268
 269