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;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["fasta"] = tempOutNames;
  32                 outputTypes["accnos"] = tempOutNames;
  33                 outputTypes["sharedseqs"] = tempOutNames;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "GetSharedOTUCommand", "GetSharedOTUCommand");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 vector<string> GetSharedOTUCommand::getRequiredParameters(){
  42         try {
  43                 string Array[] =  {"list","group"};
  44                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  45                 return myArray;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "GetSharedOTUCommand", "getRequiredParameters");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 vector<string> GetSharedOTUCommand::getRequiredFiles(){
  54         try {
  55                 vector<string> myArray;
  56                 return myArray;
  57         }
  58         catch(exception& e) {
  59                 m->errorOut(e, "GetSharedOTUCommand", "getRequiredFiles");
  60                 exit(1);
  61         }
  62 }
  63 //**********************************************************************************************************************
  64 GetSharedOTUCommand::GetSharedOTUCommand(string option)  {
  65         try {
  66
  67                 globaldata = GlobalData::getInstance();
  68                 abort = false;
  69                 unique = true;
  70                 allLines = 1;
  71                 labels.clear();
  72
  73                 //allow user to run help
  74                 if(option == "help") { help(); abort = true; }
  75
  76                 else {
  77                         //valid paramters for this command
  78                         string Array[] =  {"label","unique","shared","fasta","list","group","output","outputdir","inputdir"};
  79                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  80
  81                         OptionParser parser(option);
  82                         map<string,string> parameters = parser.getParameters();
  83
  84                         ValidParameters validParameter;
  85                         map<string,string>::iterator it;
  86
  87                         //check to make sure all parameters are valid for command
  88                         for (it = parameters.begin(); it != parameters.end(); it++) {
  89                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  90                         }
  91
  92                         //initialize outputTypes
  93                         vector<string> tempOutNames;
  94                         outputTypes["fasta"] = tempOutNames;
  95                         outputTypes["accnos"] = tempOutNames;
  96                         outputTypes["sharedseqs"] = tempOutNames;
  97
  98                         //if the user changes the output directory command factory will send this info to us in the output parameter
  99                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 100
 101                         //if the user changes the input directory command factory will send this info to us in the output parameter
 102                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 103                         if (inputDir == "not found"){   inputDir = "";          }
 104                         else {
 105                                 string path;
 106                                 it = parameters.find("fasta");
 107                                 //user has given a template file
 108                                 if(it != parameters.end()){
 109                                         path = m->hasPath(it->second);
 110                                         //if the user has not given a path then, add inputdir. else leave path alone.
 111                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 112                                 }
 113
 114                                 it = parameters.find("list");
 115                                 //user has given a template file
 116                                 if(it != parameters.end()){
 117                                         path = m->hasPath(it->second);
 118                                         //if the user has not given a path then, add inputdir. else leave path alone.
 119                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 120                                 }
 121
 122                                 it = parameters.find("group");
 123                                 //user has given a template file
 124                                 if(it != parameters.end()){
 125                                         path = m->hasPath(it->second);
 126                                         //if the user has not given a path then, add inputdir. else leave path alone.
 127                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 128                                 }
 129                         }
 130
 131
 132                         //check for required parameters
 133                         listfile = validParameter.validFile(parameters, "list", true);
 134                         if (listfile == "not open") { abort = true; }
 135                         else if (listfile == "not found") { listfile = ""; }
 136                         else {  globaldata->setListFile(listfile);  globaldata->setFormat("list");      }
 137
 138                         groupfile = validParameter.validFile(parameters, "group", true);
 139                         if (groupfile == "not open") { abort = true; }
 140                         else if (groupfile == "not found") { groupfile = ""; }
 141
 142                         if ((listfile == "") || (groupfile == "")) { m->mothurOut("The list and group parameters are required."); m->mothurOutEndLine(); abort = true; }
 143
 144                         //check for optional parameter and set defaults
 145                         // ...at some point should added some additional type checking...
 146                         label = validParameter.validFile(parameters, "label", false);
 147                         if (label == "not found") { label = ""; }
 148                         else {
 149                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 150                                 else { allLines = 1;  }
 151                         }
 152
 153                         output = validParameter.validFile(parameters, "output", false);
 154                         if (output == "not found") { output = ""; }
 155
 156                         groups = validParameter.validFile(parameters, "unique", false);
 157                         if (groups == "not found") { groups = ""; }
 158                         else {
 159                                 userGroups = "unique." + groups;
 160                                 m->splitAtDash(groups, Groups);
 161                                 globaldata->Groups = Groups;
 162
 163                         }
 164
 165                         groups = validParameter.validFile(parameters, "shared", false);
 166                         if (groups == "not found") { groups = "";  }
 167                         else {
 168                                 userGroups = groups;
 169                                 m->splitAtDash(groups, Groups);
 170                                 globaldata->Groups = Groups;
 171                                 unique = false;
 172                         }
 173
 174                         fastafile = validParameter.validFile(parameters, "fasta", true);
 175                         if (fastafile == "not open") { abort = true; }
 176                         else if (fastafile == "not found") {  fastafile = "";  }
 177
 178                 }
 179
 180         }
 181         catch(exception& e) {
 182                 m->errorOut(e, "GetSharedOTUCommand", "GetSharedOTUCommand");
 183                 exit(1);
 184         }
 185 }
 186 //**********************************************************************************************************************
 187
 188 void GetSharedOTUCommand::help(){
 189         try {
 190                 m->mothurOut("The get.sharedseqs command parameters are list, group, label, unique, shared, output and fasta.  The list and group parameters are required.\n");
 191                 m->mothurOut("The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n");
 192                 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");
 193                 m->mothurOut("If you enter your groups under the unique parameter mothur will return the otus that contain ONLY sequences from those groups.\n");
 194                 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");
 195                 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");
 196                 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");
 197                 m->mothurOut("The output parameter allows you to output the list of names without the group and bin number added. \n");
 198                 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");
 199                 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");
 200                 m->mothurOut("The get.sharedseqs command should be in the following format: get.sabund(label=yourLabels, groups=yourGroups, fasta=yourFastafile, output=yourOutput).\n");
 201                 m->mothurOut("Example get.sharedseqs(list=amazon.fn.list, label=unique-0.01, group=forest-pasture, fasta=amazon.fasta, output=accnos).\n");
 202                 m->mothurOut("The output to the screen is the distance and the number of otus at that distance for the groups you specified.\n");
 203                 m->mothurOut("The default value for label is all labels in your inputfile. The default for groups is all groups in your file.\n");
 204                 m->mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabel).\n\n");
 205         }
 206         catch(exception& e) {
 207                 m->errorOut(e, "GetSharedOTUCommand", "help");
 208                 exit(1);
 209         }
 210 }
 211
 212 //**********************************************************************************************************************
 213
 214 GetSharedOTUCommand::~GetSharedOTUCommand(){}
 215
 216 //**********************************************************************************************************************
 217
 218 int GetSharedOTUCommand::execute(){
 219         try {
 220
 221                 if (abort == true) { return 0; }
 222
 223                 groupMap = new GroupMap(groupfile);
 224                 int error = groupMap->readMap();
 225                 if (error == 1) { delete groupMap; return 0; }
 226
 227                 if (m->control_pressed) { delete groupMap; return 0; }
 228
 229                 globaldata->gGroupmap = groupMap;
 230
 231                 if (Groups.size() == 0) {
 232                         Groups = groupMap->namesOfGroups;
 233
 234                         //make string for outputfile name
 235                         userGroups = "unique.";
 236                         for(int i = 0; i < Groups.size(); i++) {  userGroups += Groups[i] + "-";  }
 237                         userGroups = userGroups.substr(0, userGroups.length()-1);
 238                 }
 239
 240                 //put groups in map to find easier
 241                 for(int i = 0; i < Groups.size(); i++) {
 242                         groupFinder[Groups[i]] = Groups[i];
 243                 }
 244
 245                 if (fastafile != "") {
 246                         ifstream inFasta;
 247                         m->openInputFile(fastafile, inFasta);
 248
 249                         while(!inFasta.eof()) {
 250                                 if (m->control_pressed) { outputTypes.clear(); inFasta.close(); delete groupMap; return 0; }
 251
 252                                 Sequence seq(inFasta); m->gobble(inFasta);
 253                                 if (seq.getName() != "") {  seqs.push_back(seq);   }
 254                         }
 255                         inFasta.close();
 256                 }
 257
 258                 ListVector* lastlist = NULL;
 259                 string lastLabel = "";
 260
 261                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 262                 set<string> processedLabels;
 263                 set<string> userLabels = labels;
 264
 265                 ifstream in;
 266                 m->openInputFile(listfile, in);
 267
 268                 //as long as you are not at the end of the file or done wih the lines you want
 269                 while((!in.eof()) && ((allLines == 1) || (userLabels.size() != 0))) {
 270
 271                         if (m->control_pressed) {
 272                                 if (lastlist != NULL) {         delete lastlist;        }
 273                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }  outputTypes.clear();
 274                                 delete groupMap; return 0;
 275                         }
 276
 277                         list = new ListVector(in);
 278
 279                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 280                                 m->mothurOut(list->getLabel());
 281                                 process(list);
 282
 283                                 processedLabels.insert(list->getLabel());
 284                                 userLabels.erase(list->getLabel());
 285                         }
 286
 287                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 288                                         string saveLabel = list->getLabel();
 289
 290                                         m->mothurOut(lastlist->getLabel());
 291                                         process(lastlist);
 292
 293                                         processedLabels.insert(lastlist->getLabel());
 294                                         userLabels.erase(lastlist->getLabel());
 295
 296                                         //restore real lastlabel to save below
 297                                         list->setLabel(saveLabel);
 298                         }
 299
 300                         lastLabel = list->getLabel();
 301
 302                         if (lastlist != NULL) {         delete lastlist;        }
 303                         lastlist = list;
 304                 }
 305
 306                 in.close();
 307
 308                 //output error messages about any remaining user labels
 309                 set<string>::iterator it;
 310                 bool needToRun = false;
 311                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 312                         m->mothurOut("Your file does not include the label " + *it);
 313                         if (processedLabels.count(lastLabel) != 1) {
 314                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 315                                 needToRun = true;
 316                         }else {
 317                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 318                         }
 319                 }
 320
 321                 //run last label if you need to
 322                 if (needToRun == true)  {
 323                                 m->mothurOut(lastlist->getLabel());
 324                                 process(lastlist);
 325
 326                                 processedLabels.insert(lastlist->getLabel());
 327                                 userLabels.erase(lastlist->getLabel());
 328                 }
 329
 330
 331                 //reset groups parameter
 332                 globaldata->Groups.clear();
 333
 334                 if (lastlist != NULL) {         delete lastlist;        }
 335
 336                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str()); }  delete groupMap; return 0; }
 337
 338                 m->mothurOutEndLine();
 339                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 340                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 341                 m->mothurOutEndLine();
 342
 343
 344                 return 0;
 345         }
 346
 347         catch(exception& e) {
 348                 m->errorOut(e, "GetSharedOTUCommand", "execute");
 349                 exit(1);
 350         }
 351 }
 352 /***********************************************************/
 353 int GetSharedOTUCommand::process(ListVector* shared) {
 354         try {
 355
 356                 map<string, string> fastaMap;
 357
 358                 ofstream outNames;
 359                 string outputFileNames;
 360
 361                 if (outputDir == "") { outputDir += m->hasPath(listfile); }
 362                 if (output != "accnos") {
 363                         outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + ".shared.seqs";
 364                 }else {
 365                         outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + ".accnos";
 366                 }
 367                 m->openOutputFile(outputFileNames, outNames);
 368
 369                 bool wroteSomething = false;
 370                 int num = 0;
 371
 372                 //go through each bin, find out if shared
 373                 for (int i = 0; i < shared->getNumBins(); i++) {
 374                         if (m->control_pressed) { outNames.close(); remove(outputFileNames.c_str()); return 0; }
 375
 376                         bool uniqueOTU = true;
 377
 378                         map<string, int> atLeastOne;
 379                         for (int f = 0; f < Groups.size(); f++) {
 380                                 atLeastOne[Groups[f]] = 0;
 381                         }
 382
 383                         vector<string> namesOfSeqsInThisBin;
 384
 385                         string names = shared->get(i);
 386                         while ((names.find_first_of(',') != -1)) {
 387                                 string name = names.substr(0,names.find_first_of(','));
 388                                 names = names.substr(names.find_first_of(',')+1, names.length());
 389
 390                                 //find group
 391                                 string seqGroup = groupMap->getGroup(name);
 392                                 if (output != "accnos") {
 393                                         namesOfSeqsInThisBin.push_back((name + "|" + seqGroup + "|" + toString(i+1)));
 394                                 }else {  namesOfSeqsInThisBin.push_back(name);  }
 395
 396                                 if (seqGroup == "not found") { m->mothurOut(name + " is not in your groupfile. Please correct."); m->mothurOutEndLine(); exit(1);  }
 397
 398                                 //is this seq in one of hte groups we care about
 399                                 it = groupFinder.find(seqGroup);
 400                                 if (it == groupFinder.end()) {  uniqueOTU = false;  } //you have a sequence from a group you don't want
 401                                 else {  atLeastOne[seqGroup]++;  }
 402                         }
 403
 404                         //get last name
 405                         string seqGroup = groupMap->getGroup(names);
 406                         if (output != "accnos") {
 407                                 namesOfSeqsInThisBin.push_back((names + "|" + seqGroup + "|" + toString(i+1)));
 408                         }else {  namesOfSeqsInThisBin.push_back(names); }
 409
 410                         if (seqGroup == "not found") { m->mothurOut(names + " is not in your groupfile. Please correct."); m->mothurOutEndLine(); exit(1);  }
 411
 412                         //is this seq in one of hte groups we care about
 413                         it = groupFinder.find(seqGroup);
 414                         if (it == groupFinder.end()) {  uniqueOTU = false;  } //you have a sequence from a group you don't want
 415                         else {  atLeastOne[seqGroup]++;  }
 416
 417
 418                         //make sure you have at least one seq from each group you want
 419                         bool sharedByAll = true;
 420                         map<string, int>::iterator it2;
 421                         for (it2 = atLeastOne.begin(); it2 != atLeastOne.end(); it2++) {
 422                                 if (it2->second == 0) {  sharedByAll = false;   }
 423                         }
 424
 425                         //if the user wants unique bins and this is unique then print
 426                         //or this the user wants shared bins and this bin is shared then print
 427                         if ((unique && uniqueOTU && sharedByAll) || (!unique && sharedByAll)) {
 428
 429                                 wroteSomething = true;
 430                                 num++;
 431
 432                                 //output list of names
 433                                 for (int j = 0; j < namesOfSeqsInThisBin.size(); j++) {
 434                                         outNames << namesOfSeqsInThisBin[j] << endl;
 435
 436                                         if (fastafile != "") {
 437                                                 if (output != "accnos") {
 438                                                         string seqName = namesOfSeqsInThisBin[j].substr(0,namesOfSeqsInThisBin[j].find_last_of('|'));
 439                                                         seqName = seqName.substr(0,seqName.find_last_of('|'));
 440                                                         fastaMap[seqName] = namesOfSeqsInThisBin[j];  //fastaMap needs to contain just the seq name for output later
 441                                                 }else {
 442                                                         fastaMap[namesOfSeqsInThisBin[j]] = namesOfSeqsInThisBin[j];
 443                                                 }
 444                                         }
 445                                 }
 446                         }
 447                 }
 448
 449                 outNames.close();
 450
 451                 if (!wroteSomething) {
 452                         remove(outputFileNames.c_str());
 453                         string outputString = "\t" + toString(num) + " - No otus shared by groups";
 454
 455                         string groupString = "";
 456                         for (int h = 0; h < Groups.size(); h++) {
 457                                 groupString += "  " + Groups[h];
 458                         }
 459
 460                         outputString += groupString + ".";
 461                         m->mothurOut(outputString); m->mothurOutEndLine();
 462                 }else {
 463                         m->mothurOut("\t" + toString(num)); m->mothurOutEndLine();
 464                         outputNames.push_back(outputFileNames);
 465                         if (output != "accnos") { outputTypes["sharedseqs"].push_back(outputFileNames); }
 466                         else { outputTypes["accnos"].push_back(outputFileNames); }
 467                 }
 468
 469                 //if fasta file provided output new fasta file
 470                 if ((fastafile != "") && wroteSomething) {
 471                         if (outputDir == "") { outputDir += m->hasPath(fastafile); }
 472                         string outputFileFasta = outputDir + m->getRootName(m->getSimpleName(fastafile)) + shared->getLabel() + userGroups + ".shared.fasta";
 473                         ofstream outFasta;
 474                         m->openOutputFile(outputFileFasta, outFasta);
 475                         outputNames.push_back(outputFileFasta); outputTypes["fasta"].push_back(outputFileFasta);
 476
 477                         for (int k = 0; k < seqs.size(); k++) {
 478                                 if (m->control_pressed) { outFasta.close(); return 0; }
 479
 480                                 //if this is a sequence we want, output it
 481                                 it = fastaMap.find(seqs[k].getName());
 482                                 if (it != fastaMap.end()) {
 483
 484                                         if (output != "accnos") {
 485                                                 outFasta << ">" << it->second << endl;
 486                                         }else {
 487                                                 outFasta << ">" << it->first << endl;
 488                                         }
 489
 490                                         outFasta << seqs[k].getAligned() << endl;
 491                                 }
 492                         }
 493
 494                         outFasta.close();
 495                 }
 496
 497                 return 0;
 498
 499         }
 500         catch(exception& e) {
 501                 m->errorOut(e, "GetSharedOTUCommand", "process");
 502                 exit(1);
 503         }
 504 }
 505
 506 //**********************************************************************************************************************