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