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