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(string option){
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16                 abort = false;
  17                 allLines = 1;
  18                 lines.clear();
  19                 labels.clear();
  20
  21                 //allow user to run help
  22                 if(option == "help") { help(); abort = true; }
  23
  24                 else {
  25                         //valid paramters for this command
  26                         string AlignArray[] =  {"fasta","line","label","name", "group"};
  27                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  28
  29                         parser = new OptionParser();
  30                         parser->parse(option, parameters);  delete parser;
  31
  32                         ValidParameters* validParameter = new ValidParameters();
  33
  34                         //check to make sure all parameters are valid for command
  35                         for (it = parameters.begin(); it != parameters.end(); it++) {
  36                                 if (validParameter->isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  37                         }
  38
  39                         //make sure the user has already run the read.otu command
  40                         if (globaldata->getListFile() == "") { cout << "You must read a listfile before running the bin.seqs command." << endl; abort = true; }
  41
  42
  43                         //check for required parameters
  44                         fastafile = validParameter->validFile(parameters, "fasta", true);
  45                         if (fastafile == "not found") { cout << "fasta is a required parameter for the bin.seqs command." << endl; abort = true; }
  46                         else if (fastafile == "not open") { abort = true; }
  47                         else {
  48                                 globaldata->setFastaFile(fastafile);
  49                                 openInputFile(fastafile, in);
  50                                 fasta = new FastaMap();
  51                         }
  52
  53
  54                         //check for optional parameter and set defaults
  55                         // ...at some point should added some additional type checking...
  56                         line = validParameter->validFile(parameters, "line", false);
  57                         if (line == "not found") { line = "";  }
  58                         else {
  59                                 if(line != "all") {  splitAtDash(line, lines);  allLines = 0;  }
  60                                 else { allLines = 1;  }
  61                         }
  62
  63                         label = validParameter->validFile(parameters, "label", false);
  64                         if (label == "not found") { label = ""; }
  65                         else {
  66                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  67                                 else { allLines = 1;  }
  68                         }
  69
  70                         //make sure user did not use both the line and label parameters
  71                         if ((line != "") && (label != "")) { cout << "You cannot use both the line and label parameters at the same time. " << endl; abort = true; }
  72                         //if the user has not specified any line or labels use the ones from read.otu
  73                         else if ((line == "") && (label == "")) {
  74                                 allLines = globaldata->allLines;
  75                                 labels = globaldata->labels;
  76                                 lines = globaldata->lines;
  77                         }
  78
  79                         namesfile = validParameter->validFile(parameters, "name", true);
  80                         if (namesfile == "not open") { abort = true; }
  81                         else if (namesfile == "not found") { namesfile = ""; }
  82
  83                         groupfile = validParameter->validFile(parameters, "group", true);
  84                         if (groupfile == "not open") { abort = true; }
  85                         else if (groupfile == "not found") { groupfile = ""; }
  86                         else {
  87                                 //read in group map info.
  88                                 groupMap = new GroupMap(groupfile);
  89                                 groupMap->readMap();
  90                         }
  91
  92                         delete validParameter;
  93                 }
  94         }
  95         catch(exception& e) {
  96                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function BinSeqCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  97                 exit(1);
  98         }
  99         catch(...) {
 100                 cout << "An unknown error has occurred in the BinSeqCommand class function BinSeqCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 101                 exit(1);
 102         }
 103 }
 104 //**********************************************************************************************************************
 105
 106 void BinSeqCommand::help(){
 107         try {
 108                 cout << "The bin.seqs command can only be executed after a successful read.otu command of a listfile." << "\n";
 109                 cout << "The bin.seqs command parameters are fasta, name, line, label and group.  The fasta parameter is required, and you may not use line and label at the same time." << "\n";
 110                 cout << "The line and label allow you to select what distance levels you would like a output files created for, and are separated by dashes." << "\n";
 111                 cout << "The bin.seqs command should be in the following format: bin.seqs(fasta=yourFastaFile, name=yourNamesFile, group=yourGroupFile, line=yourLines, label=yourLabels)." << "\n";
 112                 cout << "Example bin.seqs(fasta=amazon.fasta, group=amazon.groups, line=1-3-5, name=amazon.names)." << "\n";
 113                 cout << "The default value for line and label are all lines in your inputfile." << "\n";
 114                 cout << "The bin.seqs command outputs a .fasta file for each distance you specify appending the OTU number to each name." << "\n";
 115                 cout << "If you provide a groupfile, then it also appends the sequences group to the name." << "\n";
 116                 cout << "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile)." << "\n" << "\n";
 117         }
 118         catch(exception& e) {
 119                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 120                 exit(1);
 121         }
 122         catch(...) {
 123                 cout << "An unknown error has occurred in the BinSeqCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 124                 exit(1);
 125         }
 126 }
 127
 128
 129
 130 //**********************************************************************************************************************
 131
 132 BinSeqCommand::~BinSeqCommand(){
 133         //made new in execute
 134         if (abort == false) {
 135                 delete input;
 136                 delete read;
 137                 delete list;
 138         }
 139
 140         //made new in constructor
 141         delete fasta;
 142         if (groupfile != "") {
 143                         delete groupMap;
 144         }
 145
 146 }
 147
 148 //**********************************************************************************************************************
 149
 150 int BinSeqCommand::execute(){
 151         try {
 152
 153                 if (abort == true) {    return 0;       }
 154
 155                 int count = 1;
 156                 int error = 0;
 157
 158                 //read fastafile
 159                 fasta->readFastaFile(in);
 160
 161                 //set format to list so input can get listvector
 162                 globaldata->setFormat("list");
 163
 164                 //if user gave a namesfile then use it
 165                 if (namesfile != "") {
 166                         readNamesFile();
 167                 }
 168
 169                 //read list file
 170                 read = new ReadOTUFile(globaldata->getListFile());
 171                 read->read(&*globaldata);
 172
 173                 input = globaldata->ginput;
 174                 list = globaldata->gListVector;
 175                 ListVector* lastList = list;
 176
 177                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 178                 set<string> processedLabels;
 179                 set<string> userLabels = labels;
 180                 set<int> userLines = lines;
 181
 182
 183                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 184
 185                         if(allLines == 1 || lines.count(count) == 1 || labels.count(list->getLabel()) == 1){
 186
 187                                 error = process(list, count);
 188                                 if (error == 1) { return 0; }
 189
 190                                 processedLabels.insert(list->getLabel());
 191                                 userLabels.erase(list->getLabel());
 192                                 userLines.erase(count);
 193                         }
 194
 195                         if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastList->getLabel()) != 1)) {
 196
 197                                 error = process(lastList, count);
 198                                 if (error == 1) { return 0; }
 199
 200                                 processedLabels.insert(lastList->getLabel());
 201                                 userLabels.erase(lastList->getLabel());
 202
 203                         }
 204
 205                         if (count != 1) { delete lastList; }
 206                         lastList = list;
 207
 208                         list = input->getListVector();
 209                         count++;
 210                 }
 211
 212
 213                 //output error messages about any remaining user labels
 214                 set<string>::iterator it;
 215                 bool needToRun = false;
 216                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 217                         cout << "Your file does not include the label "<< *it;
 218                         if (processedLabels.count(lastList->getLabel()) != 1) {
 219                                 cout << ". I will use " << lastList->getLabel() << "." << endl;
 220                                 needToRun = true;
 221                         }else {
 222                                 cout << ". Please refer to " << lastList->getLabel() << "." << endl;
 223                         }
 224                 }
 225
 226                 //run last line if you need to
 227                 if (needToRun == true)  {
 228                         error = process(lastList, count);
 229                         if (error == 1) { return 0; }
 230                 }
 231
 232                 delete lastList;
 233                 return 0;
 234         }
 235         catch(exception& e) {
 236                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 237                 exit(1);
 238         }
 239         catch(...) {
 240                 cout << "An unknown error has occurred in the BinSeqCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 241                 exit(1);
 242         }
 243 }
 244
 245 //**********************************************************************************************************************
 246 void BinSeqCommand::readNamesFile() {
 247         try {
 248                 vector<string> dupNames;
 249                 openInputFile(namesfile, inNames);
 250
 251                 string name, names, sequence;
 252
 253                 while(inNames){
 254                         inNames >> name;                        //read from first column  A
 255                         inNames >> names;               //read from second column  A,B,C,D
 256
 257                         dupNames.clear();
 258
 259                         //parse names into vector
 260                         splitAtComma(names, dupNames);
 261
 262                         //store names in fasta map
 263                         sequence = fasta->getSequence(name);
 264                         for (int i = 0; i < dupNames.size(); i++) {
 265                                 fasta->push_back(dupNames[i], sequence);
 266                         }
 267
 268                         gobble(inNames);
 269                 }
 270                 inNames.close();
 271
 272         }
 273         catch(exception& e) {
 274                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 275                 exit(1);
 276         }
 277         catch(...) {
 278                 cout << "An unknown error has occurred in the BinSeqCommand class function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 279                 exit(1);
 280         }
 281 }
 282 //**********************************************************************************************************************
 283 //return 1 if error, 0 otherwise
 284 int BinSeqCommand::process(ListVector* list, int count) {
 285         try {
 286                                 string binnames, name, sequence;
 287                                 string outputFileName = getRootName(globaldata->getListFile()) + list->getLabel() + ".fasta";
 288                                 openOutputFile(outputFileName, out);
 289
 290                                 cout << list->getLabel() << '\t' << count << endl;
 291
 292                                 //for each bin in the list vector
 293                                 for (int i = 0; i < list->size(); i++) {
 294
 295                                         binnames = list->get(i);
 296                                         while (binnames.find_first_of(',') != -1) {
 297                                                 name = binnames.substr(0,binnames.find_first_of(','));
 298                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 299
 300                                                 //do work for that name
 301                                                 sequence = fasta->getSequence(name);
 302                                                 if (sequence != "not found") {
 303                                                         //if you don't have groups
 304                                                         if (groupfile == "") {
 305                                                                 name = name + "|" + toString(i+1);
 306                                                                 out << ">" << name << endl;
 307                                                                 out << sequence << endl;
 308                                                         }else {//if you do have groups
 309                                                                 string group = groupMap->getGroup(name);
 310                                                                 if (group == "not found") {
 311                                                                         cout << name << " is missing from your group file. Please correct. " << endl;
 312                                                                         remove(outputFileName.c_str());
 313                                                                         return 1;
 314                                                                 }else{
 315                                                                         name = name + "|" + group + "|" + toString(i+1);
 316                                                                         out << ">" << name << endl;
 317                                                                         out << sequence << endl;
 318                                                                 }
 319                                                         }
 320                                                 }else {
 321                                                         cout << name << " is missing from your fasta or name file. Please correct. " << endl;
 322                                                         remove(outputFileName.c_str());
 323                                                         return 1;
 324                                                 }
 325
 326                                         }
 327
 328                                         //get last name
 329                                         sequence = fasta->getSequence(binnames);
 330                                         if (sequence != "not found") {
 331                                                 //if you don't have groups
 332                                                 if (groupfile == "") {
 333                                                         binnames = binnames + "|" + toString(i+1);
 334                                                         out << ">" << binnames << endl;
 335                                                         out << sequence << endl;
 336                                                 }else {//if you do have groups
 337                                                         string group = groupMap->getGroup(binnames);
 338                                                         if (group == "not found") {
 339                                                                 cout << binnames << " is missing from your group file. Please correct. " << endl;
 340                                                                 remove(outputFileName.c_str());
 341                                                                 return 1;
 342                                                         }else{
 343                                                                 binnames = binnames + "|" + group + "|" + toString(i+1);
 344                                                                 out << ">" << binnames << endl;
 345                                                                 out << sequence << endl;
 346                                                         }
 347                                                 }
 348                                         }else {
 349                                                 cout << binnames << " is missing from your fasta or name file. Please correct. " << endl;
 350                                                 remove(outputFileName.c_str());
 351                                                 return 1;
 352                                         }
 353                                 }
 354
 355                                 out.close();
 356                                 return 0;
 357
 358         }
 359         catch(exception& e) {
 360                 cout << "Standard Error: " << e.what() << " has occurred in the BinSeqCommand class Function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 361                 exit(1);
 362         }
 363         catch(...) {
 364                 cout << "An unknown error has occurred in the BinSeqCommand class function readNamesFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 365                 exit(1);
 366         }
 367 }
 368 //**********************************************************************************************************************
 369
 370