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; calledHelp = true;
  52                 vector<string> tempOutNames;
  53                 outputTypes["fasta"] = tempOutNames;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "BinSeqCommand", "BinSeqCommand");
  57                 exit(1);
  58         }
  59 }
  60 //**********************************************************************************************************************
  61 BinSeqCommand::BinSeqCommand(string option) {
  62         try {
  63                 globaldata = GlobalData::getInstance();
  64                 abort = false; calledHelp = false;
  65                 allLines = 1;
  66                 labels.clear();
  67
  68                 //allow user to run help
  69                 if(option == "help") { help(); abort = true; calledHelp = true; }
  70
  71                 else {
  72                         //valid paramters for this command
  73                         string AlignArray[] =  {"fasta","label","name", "group","outputdir","inputdir"};
  74                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  75
  76                         OptionParser parser(option);
  77                         map<string, string> parameters = parser.getParameters();
  78
  79                         ValidParameters validParameter;
  80                         map<string, string>::iterator it;
  81
  82                         //check to make sure all parameters are valid for command
  83                         for (it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86
  87                         //initialize outputTypes
  88                         vector<string> tempOutNames;
  89                         outputTypes["fasta"] = tempOutNames;
  90
  91                         //if the user changes the output directory command factory will send this info to us in the output parameter
  92                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  93                                 outputDir = "";
  94                                 outputDir += m->hasPath(globaldata->getListFile()); //if user entered a file with a path then preserve it
  95                         }
  96
  97
  98                         //make sure the user has already run the read.otu command
  99                         if (globaldata->getListFile() == "") {
 100                                 m->mothurOut("You must read a listfile before running the bin.seqs command.");
 101                                 m->mothurOutEndLine();
 102                                 abort = true;
 103                         }
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("fasta");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 116                                 }
 117
 118                                 it = parameters.find("name");
 119                                 //user has given a template file
 120                                 if(it != parameters.end()){
 121                                         path = m->hasPath(it->second);
 122                                         //if the user has not given a path then, add inputdir. else leave path alone.
 123                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 124                                 }
 125
 126                                 it = parameters.find("group");
 127                                 //user has given a template file
 128                                 if(it != parameters.end()){
 129                                         path = m->hasPath(it->second);
 130                                         //if the user has not given a path then, add inputdir. else leave path alone.
 131                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 132                                 }
 133                         }
 134
 135
 136                         //check for required parameters
 137                         fastafile = validParameter.validFile(parameters, "fasta", true);
 138                         if (fastafile == "not found") { m->mothurOut("fasta is a required parameter for the bin.seqs command.");  m->mothurOutEndLine(); abort = true; }
 139                         else if (fastafile == "not open") { abort = true; }
 140
 141                         //check for optional parameter and set defaults
 142                         // ...at some point should added some additional type checking...
 143
 144                         label = validParameter.validFile(parameters, "label", false);
 145                         if (label == "not found") { label = ""; }
 146                         else {
 147                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 148                                 else { allLines = 1;  }
 149                         }
 150
 151                         //if the user has not specified any labels use the ones from read.otu
 152                         if (label == "") {
 153                                 allLines = globaldata->allLines;
 154                                 labels = globaldata->labels;
 155                         }
 156
 157                         namesfile = validParameter.validFile(parameters, "name", true);
 158                         if (namesfile == "not open") { abort = true; }
 159                         else if (namesfile == "not found") { namesfile = ""; }
 160
 161                         groupfile = validParameter.validFile(parameters, "group", true);
 162                         if (groupfile == "not open") { abort = true; }
 163                         else if (groupfile == "not found") { groupfile = ""; }
 164
 165                         if (abort == false) {
 166 //                              m->openInputFile(fastafile, in);
 167                                 fasta = new FastaMap();
 168                                 if (groupfile != "") {
 169                                         groupMap = new GroupMap(groupfile);
 170
 171                                         int error = groupMap->readMap();
 172                                         if (error == 1) { delete groupMap; abort = true; }
 173                                 }
 174                         }
 175
 176                 }
 177         }
 178         catch(exception& e) {
 179                 m->errorOut(e, "BinSeqCommand", "BinSeqCommand");
 180                 exit(1);
 181         }
 182 }
 183 //**********************************************************************************************************************
 184
 185 void BinSeqCommand::help(){
 186         try {
 187                 m->mothurOut("The bin.seqs command can only be executed after a successful read.otu command of a listfile.\n");
 188                 m->mothurOut("The bin.seqs command parameters are fasta, name, label and group.  The fasta parameter is required.\n");
 189                 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");
 190                 m->mothurOut("The bin.seqs command should be in the following format: bin.seqs(fasta=yourFastaFile, name=yourNamesFile, group=yourGroupFile, label=yourLabels).\n");
 191                 m->mothurOut("Example bin.seqs(fasta=amazon.fasta, group=amazon.groups, name=amazon.names).\n");
 192                 m->mothurOut("The default value for label is all lines in your inputfile.\n");
 193                 m->mothurOut("The bin.seqs command outputs a .fasta file for each distance you specify appending the OTU number to each name.\n");
 194                 m->mothurOut("If you provide a groupfile, then it also appends the sequences group to the name.\n");
 195                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 196         }
 197         catch(exception& e) {
 198                 m->errorOut(e, "BinSeqCommand", "help");
 199                 exit(1);
 200         }
 201 }
 202
 203 //**********************************************************************************************************************
 204
 205 BinSeqCommand::~BinSeqCommand(){
 206         //made new in execute
 207         if (abort == false) {
 208                 delete input;  globaldata->ginput = NULL;
 209                 delete read;
 210                 globaldata->gListVector = NULL;
 211                 delete fasta;
 212                 if (groupfile != "") {  delete groupMap;  globaldata->gGroupmap = NULL; }
 213         }
 214 }
 215
 216 //**********************************************************************************************************************
 217
 218 int BinSeqCommand::execute(){
 219         try {
 220                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 221
 222                 int error = 0;
 223
 224                 //read fastafile
 225                 fasta->readFastaFile(fastafile);
 226
 227
 228                 //set format to list so input can get listvector
 229 //              globaldata->setFormat("list");
 230
 231                 //if user gave a namesfile then use it
 232                 if (namesfile != "") {
 233                         readNamesFile();
 234                 }
 235
 236                 //read list file
 237                 read = new ReadOTUFile(globaldata->getListFile());
 238                 read->read(&*globaldata);
 239
 240                 input = globaldata->ginput;
 241                 list = globaldata->gListVector;
 242                 string lastLabel = list->getLabel();
 243
 244                 if (m->control_pressed) {  return 0; }
 245
 246                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 247                 set<string> processedLabels;
 248                 set<string> userLabels = labels;
 249
 250
 251                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 252
 253                         if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());         } return 0; }
 254
 255                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 256
 257                                 error = process(list);
 258                                 if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } return 0; }
 259
 260                                 processedLabels.insert(list->getLabel());
 261                                 userLabels.erase(list->getLabel());
 262                         }
 263
 264                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 265                                 string saveLabel = list->getLabel();
 266
 267                                 delete list;
 268                                 list = input->getListVector(lastLabel);
 269
 270                                 error = process(list);
 271                                 if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } return 0; }
 272
 273                                 processedLabels.insert(list->getLabel());
 274                                 userLabels.erase(list->getLabel());
 275
 276                                 //restore real lastlabel to save below
 277                                 list->setLabel(saveLabel);
 278                         }
 279
 280                         lastLabel = list->getLabel();
 281
 282                         delete list;
 283                         list = input->getListVector();
 284                 }
 285
 286                 if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());         } return 0; }
 287
 288                 //output error messages about any remaining user labels
 289                 set<string>::iterator it;
 290                 bool needToRun = false;
 291                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 292                         m->mothurOut("Your file does not include the label " + *it);
 293                         if (processedLabels.count(lastLabel) != 1) {
 294                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 295                                 needToRun = true;
 296                         }else {
 297                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 298                         }
 299                 }
 300
 301                 //run last label if you need to
 302                 if (needToRun == true)  {
 303                         if (list != NULL) {             delete list;    }
 304                         list = input->getListVector(lastLabel);
 305
 306                         error = process(list);
 307                         if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } return 0; }
 308
 309                         delete list;
 310                 }
 311
 312                 if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());         } return 0; }
 313
 314                 m->mothurOutEndLine();
 315                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 316                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 317                 m->mothurOutEndLine();
 318
 319
 320                 return 0;
 321         }
 322         catch(exception& e) {
 323                 m->errorOut(e, "BinSeqCommand", "execute");
 324                 exit(1);
 325         }
 326 }
 327
 328 //**********************************************************************************************************************
 329 void BinSeqCommand::readNamesFile() {
 330         try {
 331                 vector<string> dupNames;
 332                 m->openInputFile(namesfile, inNames);
 333
 334                 string name, names, sequence;
 335
 336                 while(inNames){
 337                         inNames >> name;                        //read from first column  A
 338                         inNames >> names;               //read from second column  A,B,C,D
 339
 340                         dupNames.clear();
 341
 342                         //parse names into vector
 343                         m->splitAtComma(names, dupNames);
 344
 345                         //store names in fasta map
 346                         sequence = fasta->getSequence(name);
 347                         for (int i = 0; i < dupNames.size(); i++) {
 348                                 fasta->push_back(dupNames[i], sequence);
 349                         }
 350
 351                         m->gobble(inNames);
 352                 }
 353                 inNames.close();
 354
 355         }
 356         catch(exception& e) {
 357                 m->errorOut(e, "BinSeqCommand", "readNamesFile");
 358                 exit(1);
 359         }
 360 }
 361 //**********************************************************************************************************************
 362 //return 1 if error, 0 otherwise
 363 int BinSeqCommand::process(ListVector* list) {
 364         try {
 365                                 string binnames, name, sequence;
 366
 367                                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(globaldata->getListFile())) + list->getLabel() + ".fasta";
 368                                 m->openOutputFile(outputFileName, out);
 369
 370                                 //save to output list of output file names
 371                                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 372
 373                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 374
 375                                 //for each bin in the list vector
 376                                 for (int i = 0; i < list->size(); i++) {
 377
 378                                         if (m->control_pressed) {  return 1; }
 379
 380                                         binnames = list->get(i);
 381                                         while (binnames.find_first_of(',') != -1) {
 382                                                 name = binnames.substr(0,binnames.find_first_of(','));
 383                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 384
 385                                                 //do work for that name
 386                                                 sequence = fasta->getSequence(name);
 387                                                 if (sequence != "not found") {
 388                                                         //if you don't have groups
 389                                                         if (groupfile == "") {
 390                                                                 name = name + "|" + toString(i+1);
 391                                                                 out << ">" << name << endl;
 392                                                                 out << sequence << endl;
 393                                                         }else {//if you do have groups
 394                                                                 string group = groupMap->getGroup(name);
 395                                                                 if (group == "not found") {
 396                                                                         m->mothurOut(name + " is missing from your group file. Please correct. ");  m->mothurOutEndLine();
 397                                                                         return 1;
 398                                                                 }else{
 399                                                                         name = name + "|" + group + "|" + toString(i+1);
 400                                                                         out << ">" << name << endl;
 401                                                                         out << sequence << endl;
 402                                                                 }
 403                                                         }
 404                                                 }else {
 405                                                         m->mothurOut(name + " is missing from your fasta or name file. Please correct. "); m->mothurOutEndLine();
 406                                                         return 1;
 407                                                 }
 408
 409                                         }
 410
 411                                         //get last name
 412                                         sequence = fasta->getSequence(binnames);
 413                                         if (sequence != "not found") {
 414                                                 //if you don't have groups
 415                                                 if (groupfile == "") {
 416                                                         binnames = binnames + "|" + toString(i+1);
 417                                                         out << ">" << binnames << endl;
 418                                                         out << sequence << endl;
 419                                                 }else {//if you do have groups
 420                                                         string group = groupMap->getGroup(binnames);
 421                                                         if (group == "not found") {
 422                                                                 m->mothurOut(binnames + " is missing from your group file. Please correct. "); m->mothurOutEndLine();
 423                                                                 return 1;
 424                                                         }else{
 425                                                                 binnames = binnames + "|" + group + "|" + toString(i+1);
 426                                                                 out << ">" << binnames << endl;
 427                                                                 out << sequence << endl;
 428                                                         }
 429                                                 }
 430                                         }else {
 431                                                 m->mothurOut(binnames + " is missing from your fasta or name file. Please correct. "); m->mothurOutEndLine();
 432                                                 return 1;
 433                                         }
 434                                 }
 435
 436                                 out.close();
 437                                 return 0;
 438
 439         }
 440         catch(exception& e) {
 441                 m->errorOut(e, "BinSeqCommand", "process");
 442                 exit(1);
 443         }
 444 }
 445 //**********************************************************************************************************************
 446
 447