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 vector<string> BinSeqCommand::getValidParameters(){
  14         try {
  15                 string AlignArray[] =  {"fasta","label","name", "group","outputdir","inputdir"};
  16                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  17                 return myArray;
  18         }
  19         catch(exception& e) {
  20                 m->errorOut(e, "BinSeqCommand", "getValidParameters");
  21                 exit(1);
  22         }
  23 }
  24 //**********************************************************************************************************************
  25 vector<string> BinSeqCommand::getRequiredParameters(){
  26         try {
  27                 string AlignArray[] =  {"fasta"};
  28                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  29                 return myArray;
  30         }
  31         catch(exception& e) {
  32                 m->errorOut(e, "BinSeqCommand", "getRequiredParameters");
  33                 exit(1);
  34         }
  35 }
  36 //**********************************************************************************************************************
  37 vector<string> BinSeqCommand::getRequiredFiles(){
  38         try {
  39                 string AlignArray[] =  {"list"};
  40                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  41                 return myArray;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "BinSeqCommand", "getRequiredFiles");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 BinSeqCommand::BinSeqCommand(){
  50         try {
  51                 abort = true;
  52                 //initialize outputTypes
  53                 vector<string> tempOutNames;
  54                 outputTypes["fasta"] = tempOutNames;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "BinSeqCommand", "BinSeqCommand");
  58                 exit(1);
  59         }
  60 }
  61 //**********************************************************************************************************************
  62 BinSeqCommand::BinSeqCommand(string option) {
  63         try {
  64                 globaldata = GlobalData::getInstance();
  65                 abort = false;
  66                 allLines = 1;
  67                 labels.clear();
  68
  69                 //allow user to run help
  70                 if(option == "help") { help(); abort = true; }
  71
  72                 else {
  73                         //valid paramters for this command
  74                         string AlignArray[] =  {"fasta","label","name", "group","outputdir","inputdir"};
  75                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  76
  77                         OptionParser parser(option);
  78                         map<string, string> parameters = parser.getParameters();
  79
  80                         ValidParameters validParameter;
  81                         map<string, string>::iterator it;
  82
  83                         //check to make sure all parameters are valid for command
  84                         for (it = parameters.begin(); it != parameters.end(); it++) {
  85                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  86                         }
  87
  88                         //initialize outputTypes
  89                         vector<string> tempOutNames;
  90                         outputTypes["fasta"] = tempOutNames;
  91
  92                         //if the user changes the output directory command factory will send this info to us in the output parameter
  93                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  94                                 outputDir = "";
  95                                 outputDir += m->hasPath(globaldata->getListFile()); //if user entered a file with a path then preserve it
  96                         }
  97
  98
  99                         //make sure the user has already run the read.otu command
 100                         if (globaldata->getListFile() == "") {
 101                                 m->mothurOut("You must read a listfile before running the bin.seqs command.");
 102                                 m->mothurOutEndLine();
 103                                 abort = true;
 104                         }
 105
 106                         //if the user changes the input directory command factory will send this info to us in the output parameter
 107                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 108                         if (inputDir == "not found"){   inputDir = "";          }
 109                         else {
 110                                 string path;
 111                                 it = parameters.find("fasta");
 112                                 //user has given a template file
 113                                 if(it != parameters.end()){
 114                                         path = m->hasPath(it->second);
 115                                         //if the user has not given a path then, add inputdir. else leave path alone.
 116                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 117                                 }
 118
 119                                 it = parameters.find("name");
 120                                 //user has given a template file
 121                                 if(it != parameters.end()){
 122                                         path = m->hasPath(it->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 125                                 }
 126
 127                                 it = parameters.find("group");
 128                                 //user has given a template file
 129                                 if(it != parameters.end()){
 130                                         path = m->hasPath(it->second);
 131                                         //if the user has not given a path then, add inputdir. else leave path alone.
 132                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 133                                 }
 134                         }
 135
 136
 137                         //check for required parameters
 138                         fastafile = validParameter.validFile(parameters, "fasta", true);
 139                         if (fastafile == "not found") { m->mothurOut("fasta is a required parameter for the bin.seqs command.");  m->mothurOutEndLine(); abort = true; }
 140                         else if (fastafile == "not open") { abort = true; }
 141
 142                         //check for optional parameter and set defaults
 143                         // ...at some point should added some additional type checking...
 144
 145                         label = validParameter.validFile(parameters, "label", false);
 146                         if (label == "not found") { label = ""; }
 147                         else {
 148                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 149                                 else { allLines = 1;  }
 150                         }
 151
 152                         //if the user has not specified any labels use the ones from read.otu
 153                         if (label == "") {
 154                                 allLines = globaldata->allLines;
 155                                 labels = globaldata->labels;
 156                         }
 157
 158                         namesfile = validParameter.validFile(parameters, "name", true);
 159                         if (namesfile == "not open") { abort = true; }
 160                         else if (namesfile == "not found") { namesfile = ""; }
 161
 162                         groupfile = validParameter.validFile(parameters, "group", true);
 163                         if (groupfile == "not open") { abort = true; }
 164                         else if (groupfile == "not found") { groupfile = ""; }
 165
 166                         if (abort == false) {
 167 //                              m->openInputFile(fastafile, in);
 168                                 fasta = new FastaMap();
 169                                 if (groupfile != "") {
 170                                         groupMap = new GroupMap(groupfile);
 171
 172                                         int error = groupMap->readMap();
 173                                         if (error == 1) { delete groupMap; abort = true; }
 174                                 }
 175                         }
 176
 177                 }
 178         }
 179         catch(exception& e) {
 180                 m->errorOut(e, "BinSeqCommand", "BinSeqCommand");
 181                 exit(1);
 182         }
 183 }
 184 //**********************************************************************************************************************
 185
 186 void BinSeqCommand::help(){
 187         try {
 188                 m->mothurOut("The bin.seqs command can only be executed after a successful read.otu command of a listfile.\n");
 189                 m->mothurOut("The bin.seqs command parameters are fasta, name, label and group.  The fasta parameter is required.\n");
 190                 m->mothurOut("The label parameter allows you to select what distance levels you would like a output files created for, and are separated by dashes.\n");
 191                 m->mothurOut("The bin.seqs command should be in the following format: bin.seqs(fasta=yourFastaFile, name=yourNamesFile, group=yourGroupFile, label=yourLabels).\n");
 192                 m->mothurOut("Example bin.seqs(fasta=amazon.fasta, group=amazon.groups, name=amazon.names).\n");
 193                 m->mothurOut("The default value for label is all lines in your inputfile.\n");
 194                 m->mothurOut("The bin.seqs command outputs a .fasta file for each distance you specify appending the OTU number to each name.\n");
 195                 m->mothurOut("If you provide a groupfile, then it also appends the sequences group to the name.\n");
 196                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 197         }
 198         catch(exception& e) {
 199                 m->errorOut(e, "BinSeqCommand", "help");
 200                 exit(1);
 201         }
 202 }
 203
 204 //**********************************************************************************************************************
 205
 206 BinSeqCommand::~BinSeqCommand(){
 207         //made new in execute
 208         if (abort == false) {
 209                 delete input;  globaldata->ginput = NULL;
 210                 delete read;
 211                 globaldata->gListVector = NULL;
 212                 delete fasta;
 213                 if (groupfile != "") {  delete groupMap;  globaldata->gGroupmap = NULL; }
 214         }
 215 }
 216
 217 //**********************************************************************************************************************
 218
 219 int BinSeqCommand::execute(){
 220         try {
 221                 if (abort == true) {    return 0;       }
 222
 223                 int error = 0;
 224
 225                 //read fastafile
 226                 fasta->readFastaFile(fastafile);
 227
 228
 229                 //set format to list so input can get listvector
 230 //              globaldata->setFormat("list");
 231
 232                 //if user gave a namesfile then use it
 233                 if (namesfile != "") {
 234                         readNamesFile();
 235                 }
 236
 237                 //read list file
 238                 read = new ReadOTUFile(globaldata->getListFile());
 239                 read->read(&*globaldata);
 240
 241                 input = globaldata->ginput;
 242                 list = globaldata->gListVector;
 243                 string lastLabel = list->getLabel();
 244
 245                 if (m->control_pressed) {  return 0; }
 246
 247                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 248                 set<string> processedLabels;
 249                 set<string> userLabels = labels;
 250
 251
 252                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 253
 254                         if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());         } return 0; }
 255
 256                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 257
 258                                 error = process(list);
 259                                 if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } return 0; }
 260
 261                                 processedLabels.insert(list->getLabel());
 262                                 userLabels.erase(list->getLabel());
 263                         }
 264
 265                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 266                                 string saveLabel = list->getLabel();
 267
 268                                 delete list;
 269                                 list = input->getListVector(lastLabel);
 270
 271                                 error = process(list);
 272                                 if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } return 0; }
 273
 274                                 processedLabels.insert(list->getLabel());
 275                                 userLabels.erase(list->getLabel());
 276
 277                                 //restore real lastlabel to save below
 278                                 list->setLabel(saveLabel);
 279                         }
 280
 281                         lastLabel = list->getLabel();
 282
 283                         delete list;
 284                         list = input->getListVector();
 285                 }
 286
 287                 if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());         } return 0; }
 288
 289                 //output error messages about any remaining user labels
 290                 set<string>::iterator it;
 291                 bool needToRun = false;
 292                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 293                         m->mothurOut("Your file does not include the label " + *it);
 294                         if (processedLabels.count(lastLabel) != 1) {
 295                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 296                                 needToRun = true;
 297                         }else {
 298                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 299                         }
 300                 }
 301
 302                 //run last label if you need to
 303                 if (needToRun == true)  {
 304                         if (list != NULL) {             delete list;    }
 305                         list = input->getListVector(lastLabel);
 306
 307                         error = process(list);
 308                         if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } return 0; }
 309
 310                         delete list;
 311                 }
 312
 313                 if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());         } return 0; }
 314
 315                 m->mothurOutEndLine();
 316                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 317                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 318                 m->mothurOutEndLine();
 319
 320
 321                 return 0;
 322         }
 323         catch(exception& e) {
 324                 m->errorOut(e, "BinSeqCommand", "execute");
 325                 exit(1);
 326         }
 327 }
 328
 329 //**********************************************************************************************************************
 330 void BinSeqCommand::readNamesFile() {
 331         try {
 332                 vector<string> dupNames;
 333                 m->openInputFile(namesfile, inNames);
 334
 335                 string name, names, sequence;
 336
 337                 while(inNames){
 338                         inNames >> name;                        //read from first column  A
 339                         inNames >> names;               //read from second column  A,B,C,D
 340
 341                         dupNames.clear();
 342
 343                         //parse names into vector
 344                         m->splitAtComma(names, dupNames);
 345
 346                         //store names in fasta map
 347                         sequence = fasta->getSequence(name);
 348                         for (int i = 0; i < dupNames.size(); i++) {
 349                                 fasta->push_back(dupNames[i], sequence);
 350                         }
 351
 352                         m->gobble(inNames);
 353                 }
 354                 inNames.close();
 355
 356         }
 357         catch(exception& e) {
 358                 m->errorOut(e, "BinSeqCommand", "readNamesFile");
 359                 exit(1);
 360         }
 361 }
 362 //**********************************************************************************************************************
 363 //return 1 if error, 0 otherwise
 364 int BinSeqCommand::process(ListVector* list) {
 365         try {
 366                                 string binnames, name, sequence;
 367
 368                                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(globaldata->getListFile())) + list->getLabel() + ".fasta";
 369                                 m->openOutputFile(outputFileName, out);
 370
 371                                 //save to output list of output file names
 372                                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 373
 374                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 375
 376                                 //for each bin in the list vector
 377                                 for (int i = 0; i < list->size(); i++) {
 378
 379                                         if (m->control_pressed) {  return 1; }
 380
 381                                         binnames = list->get(i);
 382                                         while (binnames.find_first_of(',') != -1) {
 383                                                 name = binnames.substr(0,binnames.find_first_of(','));
 384                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 385
 386                                                 //do work for that name
 387                                                 sequence = fasta->getSequence(name);
 388                                                 if (sequence != "not found") {
 389                                                         //if you don't have groups
 390                                                         if (groupfile == "") {
 391                                                                 name = name + "|" + toString(i+1);
 392                                                                 out << ">" << name << endl;
 393                                                                 out << sequence << endl;
 394                                                         }else {//if you do have groups
 395                                                                 string group = groupMap->getGroup(name);
 396                                                                 if (group == "not found") {
 397                                                                         m->mothurOut(name + " is missing from your group file. Please correct. ");  m->mothurOutEndLine();
 398                                                                         return 1;
 399                                                                 }else{
 400                                                                         name = name + "|" + group + "|" + toString(i+1);
 401                                                                         out << ">" << name << endl;
 402                                                                         out << sequence << endl;
 403                                                                 }
 404                                                         }
 405                                                 }else {
 406                                                         m->mothurOut(name + " is missing from your fasta or name file. Please correct. "); m->mothurOutEndLine();
 407                                                         return 1;
 408                                                 }
 409
 410                                         }
 411
 412                                         //get last name
 413                                         sequence = fasta->getSequence(binnames);
 414                                         if (sequence != "not found") {
 415                                                 //if you don't have groups
 416                                                 if (groupfile == "") {
 417                                                         binnames = binnames + "|" + toString(i+1);
 418                                                         out << ">" << binnames << endl;
 419                                                         out << sequence << endl;
 420                                                 }else {//if you do have groups
 421                                                         string group = groupMap->getGroup(binnames);
 422                                                         if (group == "not found") {
 423                                                                 m->mothurOut(binnames + " is missing from your group file. Please correct. "); m->mothurOutEndLine();
 424                                                                 return 1;
 425                                                         }else{
 426                                                                 binnames = binnames + "|" + group + "|" + toString(i+1);
 427                                                                 out << ">" << binnames << endl;
 428                                                                 out << sequence << endl;
 429                                                         }
 430                                                 }
 431                                         }else {
 432                                                 m->mothurOut(binnames + " is missing from your fasta or name file. Please correct. "); m->mothurOutEndLine();
 433                                                 return 1;
 434                                         }
 435                                 }
 436
 437                                 out.close();
 438                                 return 0;
 439
 440         }
 441         catch(exception& e) {
 442                 m->errorOut(e, "BinSeqCommand", "process");
 443                 exit(1);
 444         }
 445 }
 446 //**********************************************************************************************************************
 447
 448