getsharedotucommand.cpp

   1 /*
   2  *  getsharedotucommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/22/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "getsharedotucommand.h"
  11
  12
  13 //**********************************************************************************************************************
  14 vector<string> GetSharedOTUCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"label","unique","shared","fasta","list","group","output","outputdir","inputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "GetSharedOTUCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 GetSharedOTUCommand::GetSharedOTUCommand(){
  27         try {
  28                 abort = true; calledHelp = true;
  29                 vector<string> tempOutNames;
  30                 outputTypes["fasta"] = tempOutNames;
  31                 outputTypes["accnos"] = tempOutNames;
  32                 outputTypes["sharedseqs"] = tempOutNames;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "GetSharedOTUCommand", "GetSharedOTUCommand");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 vector<string> GetSharedOTUCommand::getRequiredParameters(){
  41         try {
  42                 string Array[] =  {"list","group"};
  43                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "GetSharedOTUCommand", "getRequiredParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 vector<string> GetSharedOTUCommand::getRequiredFiles(){
  53         try {
  54                 vector<string> myArray;
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "GetSharedOTUCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63 GetSharedOTUCommand::GetSharedOTUCommand(string option)  {
  64         try {
  65
  66                 globaldata = GlobalData::getInstance();
  67                 abort = false; calledHelp = false;
  68                 unique = true;
  69                 allLines = 1;
  70                 labels.clear();
  71
  72                 //allow user to run help
  73                 if(option == "help") { help(); abort = true; calledHelp = true; }
  74
  75                 else {
  76                         //valid paramters for this command
  77                         string Array[] =  {"label","unique","shared","fasta","list","group","output","outputdir","inputdir"};
  78                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  79
  80                         OptionParser parser(option);
  81                         map<string,string> parameters = parser.getParameters();
  82
  83                         ValidParameters validParameter;
  84                         map<string,string>::iterator it;
  85
  86                         //check to make sure all parameters are valid for command
  87                         for (it = parameters.begin(); it != parameters.end(); it++) {
  88                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  89                         }
  90
  91                         //initialize outputTypes
  92                         vector<string> tempOutNames;
  93                         outputTypes["fasta"] = tempOutNames;
  94                         outputTypes["accnos"] = tempOutNames;
  95                         outputTypes["sharedseqs"] = tempOutNames;
  96
  97                         //if the user changes the output directory command factory will send this info to us in the output parameter
  98                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  99
 100                         //if the user changes the input directory command factory will send this info to us in the output parameter
 101                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 102                         if (inputDir == "not found"){   inputDir = "";          }
 103                         else {
 104                                 string path;
 105                                 it = parameters.find("fasta");
 106                                 //user has given a template file
 107                                 if(it != parameters.end()){
 108                                         path = m->hasPath(it->second);
 109                                         //if the user has not given a path then, add inputdir. else leave path alone.
 110                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 111                                 }
 112
 113                                 it = parameters.find("list");
 114                                 //user has given a template file
 115                                 if(it != parameters.end()){
 116                                         path = m->hasPath(it->second);
 117                                         //if the user has not given a path then, add inputdir. else leave path alone.
 118                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 119                                 }
 120
 121                                 it = parameters.find("group");
 122                                 //user has given a template file
 123                                 if(it != parameters.end()){
 124                                         path = m->hasPath(it->second);
 125                                         //if the user has not given a path then, add inputdir. else leave path alone.
 126                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 127                                 }
 128                         }
 129
 130
 131                         //check for required parameters
 132                         listfile = validParameter.validFile(parameters, "list", true);
 133                         if (listfile == "not open") { abort = true; }
 134                         else if (listfile == "not found") { listfile = ""; }
 135                         else {  globaldata->setListFile(listfile);  globaldata->setFormat("list");      }
 136
 137                         groupfile = validParameter.validFile(parameters, "group", true);
 138                         if (groupfile == "not open") { abort = true; }
 139                         else if (groupfile == "not found") { groupfile = ""; }
 140
 141                         if ((listfile == "") || (groupfile == "")) { m->mothurOut("The list and group parameters are required."); m->mothurOutEndLine(); abort = true; }
 142
 143                         //check for optional parameter and set defaults
 144                         // ...at some point should added some additional type checking...
 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                         output = validParameter.validFile(parameters, "output", false);
 153                         if (output == "not found") { output = ""; }
 154
 155                         groups = validParameter.validFile(parameters, "unique", false);
 156                         if (groups == "not found") { groups = ""; }
 157                         else {
 158                                 userGroups = "unique." + groups;
 159                                 m->splitAtDash(groups, Groups);
 160                                 globaldata->Groups = Groups;
 161
 162                         }
 163
 164                         groups = validParameter.validFile(parameters, "shared", false);
 165                         if (groups == "not found") { groups = "";  }
 166                         else {
 167                                 userGroups = groups;
 168                                 m->splitAtDash(groups, Groups);
 169                                 globaldata->Groups = Groups;
 170                                 unique = false;
 171                         }
 172
 173                         fastafile = validParameter.validFile(parameters, "fasta", true);
 174                         if (fastafile == "not open") { abort = true; }
 175                         else if (fastafile == "not found") {  fastafile = "";  }
 176
 177                 }
 178
 179         }
 180         catch(exception& e) {
 181                 m->errorOut(e, "GetSharedOTUCommand", "GetSharedOTUCommand");
 182                 exit(1);
 183         }
 184 }
 185 //**********************************************************************************************************************
 186
 187 void GetSharedOTUCommand::help(){
 188         try {
 189                 m->mothurOut("The get.sharedseqs command parameters are list, group, label, unique, shared, output and fasta.  The list and group parameters are required.\n");
 190                 m->mothurOut("The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n");
 191                 m->mothurOut("The unique and shared parameters allow you to select groups you would like to know the shared info for, and are separated by dashes.\n");
 192                 m->mothurOut("If you enter your groups under the unique parameter mothur will return the otus that contain ONLY sequences from those groups.\n");
 193                 m->mothurOut("If you enter your groups under the shared parameter mothur will return the otus that contain sequences from those groups and may also contain sequences from other groups.\n");
 194                 m->mothurOut("If you do not enter any groups then the get.sharedseqs command will return sequences that are unique to all groups in your group file.\n");
 195                 m->mothurOut("The fasta parameter allows you to input a fasta file and outputs a fasta file for each distance level containing only the sequences that are in OTUs shared by the groups specified.\n");
 196                 m->mothurOut("The output parameter allows you to output the list of names without the group and bin number added. \n");
 197                 m->mothurOut("With this option you can use the names file as an input in get.seqs and remove.seqs commands. To do this enter output=accnos. \n");
 198                 m->mothurOut("The get.sharedseqs command outputs a .names file for each distance level containing a list of sequences in the OTUs shared by the groups specified.\n");
 199                 m->mothurOut("The get.sharedseqs command should be in the following format: get.sharedseqs(label=yourLabels, groups=yourGroups, fasta=yourFastafile, output=yourOutput).\n");
 200                 m->mothurOut("Example get.sharedseqs(list=amazon.fn.list, label=unique-0.01, group=forest-pasture, fasta=amazon.fasta, output=accnos).\n");
 201                 m->mothurOut("The output to the screen is the distance and the number of otus at that distance for the groups you specified.\n");
 202                 m->mothurOut("The default value for label is all labels in your inputfile. The default for groups is all groups in your file.\n");
 203                 m->mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabel).\n\n");
 204         }
 205         catch(exception& e) {
 206                 m->errorOut(e, "GetSharedOTUCommand", "help");
 207                 exit(1);
 208         }
 209 }
 210
 211 //**********************************************************************************************************************
 212
 213 GetSharedOTUCommand::~GetSharedOTUCommand(){}
 214
 215 //**********************************************************************************************************************
 216
 217 int GetSharedOTUCommand::execute(){
 218         try {
 219
 220                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 221
 222                 groupMap = new GroupMap(groupfile);
 223                 int error = groupMap->readMap();
 224                 if (error == 1) { delete groupMap; return 0; }
 225
 226                 if (m->control_pressed) { delete groupMap; return 0; }
 227
 228                 globaldata->gGroupmap = groupMap;
 229
 230                 if (Groups.size() == 0) {
 231                         Groups = groupMap->namesOfGroups;
 232
 233                         //make string for outputfile name
 234                         userGroups = "unique.";
 235                         for(int i = 0; i < Groups.size(); i++) {  userGroups += Groups[i] + "-";  }
 236                         userGroups = userGroups.substr(0, userGroups.length()-1);
 237                 }
 238
 239                 //put groups in map to find easier
 240                 for(int i = 0; i < Groups.size(); i++) {
 241                         groupFinder[Groups[i]] = Groups[i];
 242                 }
 243
 244                 if (fastafile != "") {
 245                         ifstream inFasta;
 246                         m->openInputFile(fastafile, inFasta);
 247
 248                         while(!inFasta.eof()) {
 249                                 if (m->control_pressed) { outputTypes.clear(); inFasta.close(); delete groupMap; return 0; }
 250
 251                                 Sequence seq(inFasta); m->gobble(inFasta);
 252                                 if (seq.getName() != "") {  seqs.push_back(seq);   }
 253                         }
 254                         inFasta.close();
 255                 }
 256
 257                 ListVector* lastlist = NULL;
 258                 string lastLabel = "";
 259
 260                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 261                 set<string> processedLabels;
 262                 set<string> userLabels = labels;
 263
 264                 ifstream in;
 265                 m->openInputFile(listfile, in);
 266
 267                 //as long as you are not at the end of the file or done wih the lines you want
 268                 while((!in.eof()) && ((allLines == 1) || (userLabels.size() != 0))) {
 269
 270                         if (m->control_pressed) {
 271                                 if (lastlist != NULL) {         delete lastlist;        }
 272                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }  outputTypes.clear();
 273                                 delete groupMap; return 0;
 274                         }
 275
 276                         list = new ListVector(in);
 277
 278                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 279                                 m->mothurOut(list->getLabel());
 280                                 process(list);
 281
 282                                 processedLabels.insert(list->getLabel());
 283                                 userLabels.erase(list->getLabel());
 284                         }
 285
 286                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 287                                         string saveLabel = list->getLabel();
 288
 289                                         m->mothurOut(lastlist->getLabel());
 290                                         process(lastlist);
 291
 292                                         processedLabels.insert(lastlist->getLabel());
 293                                         userLabels.erase(lastlist->getLabel());
 294
 295                                         //restore real lastlabel to save below
 296                                         list->setLabel(saveLabel);
 297                         }
 298
 299                         lastLabel = list->getLabel();
 300
 301                         if (lastlist != NULL) {         delete lastlist;        }
 302                         lastlist = list;
 303                 }
 304
 305                 in.close();
 306
 307                 //output error messages about any remaining user labels
 308                 set<string>::iterator it;
 309                 bool needToRun = false;
 310                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 311                         m->mothurOut("Your file does not include the label " + *it);
 312                         if (processedLabels.count(lastLabel) != 1) {
 313                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 314                                 needToRun = true;
 315                         }else {
 316                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 317                         }
 318                 }
 319
 320                 //run last label if you need to
 321                 if (needToRun == true)  {
 322                                 m->mothurOut(lastlist->getLabel());
 323                                 process(lastlist);
 324
 325                                 processedLabels.insert(lastlist->getLabel());
 326                                 userLabels.erase(lastlist->getLabel());
 327                 }
 328
 329
 330                 //reset groups parameter
 331                 globaldata->Groups.clear();
 332
 333                 if (lastlist != NULL) {         delete lastlist;        }
 334
 335                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str()); }  delete groupMap; return 0; }
 336
 337                 m->mothurOutEndLine();
 338                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 339                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 340                 m->mothurOutEndLine();
 341
 342
 343                 return 0;
 344         }
 345
 346         catch(exception& e) {
 347                 m->errorOut(e, "GetSharedOTUCommand", "execute");
 348                 exit(1);
 349         }
 350 }
 351 /***********************************************************/
 352 int GetSharedOTUCommand::process(ListVector* shared) {
 353         try {
 354
 355                 map<string, string> fastaMap;
 356
 357                 ofstream outNames;
 358                 string outputFileNames;
 359
 360                 if (outputDir == "") { outputDir += m->hasPath(listfile); }
 361                 if (output != "accnos") {
 362                         outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + ".shared.seqs";
 363                 }else {
 364                         outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + ".accnos";
 365                 }
 366                 m->openOutputFile(outputFileNames, outNames);
 367
 368                 bool wroteSomething = false;
 369                 int num = 0;
 370
 371                 //go through each bin, find out if shared
 372                 for (int i = 0; i < shared->getNumBins(); i++) {
 373                         if (m->control_pressed) { outNames.close(); remove(outputFileNames.c_str()); return 0; }
 374
 375                         bool uniqueOTU = true;
 376
 377                         map<string, int> atLeastOne;
 378                         for (int f = 0; f < Groups.size(); f++) {
 379                                 atLeastOne[Groups[f]] = 0;
 380                         }
 381
 382                         vector<string> namesOfSeqsInThisBin;
 383
 384                         string names = shared->get(i);
 385                         while ((names.find_first_of(',') != -1)) {
 386                                 string name = names.substr(0,names.find_first_of(','));
 387                                 names = names.substr(names.find_first_of(',')+1, names.length());
 388
 389                                 //find group
 390                                 string seqGroup = groupMap->getGroup(name);
 391                                 if (output != "accnos") {
 392                                         namesOfSeqsInThisBin.push_back((name + "|" + seqGroup + "|" + toString(i+1)));
 393                                 }else {  namesOfSeqsInThisBin.push_back(name);  }
 394
 395                                 if (seqGroup == "not found") { m->mothurOut(name + " is not in your groupfile. Please correct."); m->mothurOutEndLine(); exit(1);  }
 396
 397                                 //is this seq in one of hte groups we care about
 398                                 it = groupFinder.find(seqGroup);
 399                                 if (it == groupFinder.end()) {  uniqueOTU = false;  } //you have a sequence from a group you don't want
 400                                 else {  atLeastOne[seqGroup]++;  }
 401                         }
 402
 403                         //get last name
 404                         string seqGroup = groupMap->getGroup(names);
 405                         if (output != "accnos") {
 406                                 namesOfSeqsInThisBin.push_back((names + "|" + seqGroup + "|" + toString(i+1)));
 407                         }else {  namesOfSeqsInThisBin.push_back(names); }
 408
 409                         if (seqGroup == "not found") { m->mothurOut(names + " is not in your groupfile. Please correct."); m->mothurOutEndLine(); exit(1);  }
 410
 411                         //is this seq in one of hte groups we care about
 412                         it = groupFinder.find(seqGroup);
 413                         if (it == groupFinder.end()) {  uniqueOTU = false;  } //you have a sequence from a group you don't want
 414                         else {  atLeastOne[seqGroup]++;  }
 415
 416
 417                         //make sure you have at least one seq from each group you want
 418                         bool sharedByAll = true;
 419                         map<string, int>::iterator it2;
 420                         for (it2 = atLeastOne.begin(); it2 != atLeastOne.end(); it2++) {
 421                                 if (it2->second == 0) {  sharedByAll = false;   }
 422                         }
 423
 424                         //if the user wants unique bins and this is unique then print
 425                         //or this the user wants shared bins and this bin is shared then print
 426                         if ((unique && uniqueOTU && sharedByAll) || (!unique && sharedByAll)) {
 427
 428                                 wroteSomething = true;
 429                                 num++;
 430
 431                                 //output list of names
 432                                 for (int j = 0; j < namesOfSeqsInThisBin.size(); j++) {
 433                                         outNames << namesOfSeqsInThisBin[j] << endl;
 434
 435                                         if (fastafile != "") {
 436                                                 if (output != "accnos") {
 437                                                         string seqName = namesOfSeqsInThisBin[j].substr(0,namesOfSeqsInThisBin[j].find_last_of('|'));
 438                                                         seqName = seqName.substr(0,seqName.find_last_of('|'));
 439                                                         fastaMap[seqName] = namesOfSeqsInThisBin[j];  //fastaMap needs to contain just the seq name for output later
 440                                                 }else {
 441                                                         fastaMap[namesOfSeqsInThisBin[j]] = namesOfSeqsInThisBin[j];
 442                                                 }
 443                                         }
 444                                 }
 445                         }
 446                 }
 447
 448                 outNames.close();
 449
 450                 if (!wroteSomething) {
 451                         remove(outputFileNames.c_str());
 452                         string outputString = "\t" + toString(num) + " - No otus shared by groups";
 453
 454                         string groupString = "";
 455                         for (int h = 0; h < Groups.size(); h++) {
 456                                 groupString += "  " + Groups[h];
 457                         }
 458
 459                         outputString += groupString + ".";
 460                         m->mothurOut(outputString); m->mothurOutEndLine();
 461                 }else {
 462                         m->mothurOut("\t" + toString(num)); m->mothurOutEndLine();
 463                         outputNames.push_back(outputFileNames);
 464                         if (output != "accnos") { outputTypes["sharedseqs"].push_back(outputFileNames); }
 465                         else { outputTypes["accnos"].push_back(outputFileNames); }
 466                 }
 467
 468                 //if fasta file provided output new fasta file
 469                 if ((fastafile != "") && wroteSomething) {
 470                         if (outputDir == "") { outputDir += m->hasPath(fastafile); }
 471                         string outputFileFasta = outputDir + m->getRootName(m->getSimpleName(fastafile)) + shared->getLabel() + userGroups + ".shared.fasta";
 472                         ofstream outFasta;
 473                         m->openOutputFile(outputFileFasta, outFasta);
 474                         outputNames.push_back(outputFileFasta); outputTypes["fasta"].push_back(outputFileFasta);
 475
 476                         for (int k = 0; k < seqs.size(); k++) {
 477                                 if (m->control_pressed) { outFasta.close(); return 0; }
 478
 479                                 //if this is a sequence we want, output it
 480                                 it = fastaMap.find(seqs[k].getName());
 481                                 if (it != fastaMap.end()) {
 482
 483                                         if (output != "accnos") {
 484                                                 outFasta << ">" << it->second << endl;
 485                                         }else {
 486                                                 outFasta << ">" << it->first << endl;
 487                                         }
 488
 489                                         outFasta << seqs[k].getAligned() << endl;
 490                                 }
 491                         }
 492
 493                         outFasta.close();
 494                 }
 495
 496                 return 0;
 497
 498         }
 499         catch(exception& e) {
 500                 m->errorOut(e, "GetSharedOTUCommand", "process");
 501                 exit(1);
 502         }
 503 }
 504
 505 //**********************************************************************************************************************