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 #include "sharedutilities.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> GetSharedOTUCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pfasta);
  17                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT", "none",false,true); parameters.push_back(pgroup);
  18                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,true); parameters.push_back(plist);
  19                 CommandParameter poutput("output", "Multiple", "accnos-default", "default", "", "", "",false,false); parameters.push_back(poutput);
  20                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  21                 CommandParameter punique("unique", "String", "", "", "", "", "",false,false); parameters.push_back(punique);
  22                 CommandParameter pshared("shared", "String", "", "", "", "", "",false,false); parameters.push_back(pshared);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "GetSharedOTUCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string GetSharedOTUCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The get.sharedseqs command parameters are list, group, label, unique, shared, output and fasta.  The list and group parameters are required, unless you have valid current files.\n";
  40                 helpString += "The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n";
  41                 helpString += "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";
  42                 helpString += "If you enter your groups under the unique parameter mothur will return the otus that contain ONLY sequences from those groups.\n";
  43                 helpString += "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";
  44                 helpString += "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";
  45                 helpString += "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";
  46                 helpString += "The output parameter allows you to output the list of names without the group and bin number added. \n";
  47                 helpString += "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";
  48                 helpString += "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";
  49                 helpString += "The get.sharedseqs command should be in the following format: get.sharedseqs(list=yourListFile, group=yourGroupFile, label=yourLabels, unique=yourGroups, fasta=yourFastafile, output=yourOutput).\n";
  50                 helpString += "Example get.sharedseqs(list=amazon.fn.list, label=unique-0.01, group= amazon.groups, unique=forest-pasture, fasta=amazon.fasta, output=accnos).\n";
  51                 helpString += "The output to the screen is the distance and the number of otus at that distance for the groups you specified.\n";
  52                 helpString += "The default value for label is all labels in your inputfile. The default for groups is all groups in your file.\n";
  53                 helpString += "Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabel).\n";
  54                 return helpString;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "GetSharedOTUCommand", "getHelpString");
  58                 exit(1);
  59         }
  60 }
  61 //**********************************************************************************************************************
  62 string GetSharedOTUCommand::getOutputFileNameTag(string type, string inputName=""){
  63         try {
  64         string outputFileName = "";
  65                 map<string, vector<string> >::iterator it;
  66
  67         //is this a type this command creates
  68         it = outputTypes.find(type);
  69         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  70         else {
  71             if (type == "fasta")            {   outputFileName =  "shared.fasta";   }
  72             else if (type == "accnos")      {   outputFileName =  "accnos";         }
  73             else if (type == "sharedseqs")  {   outputFileName =  "shared.seqs";    }
  74             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  75         }
  76         return outputFileName;
  77         }
  78         catch(exception& e) {
  79                 m->errorOut(e, "GetSharedOTUCommand", "getOutputFileNameTag");
  80                 exit(1);
  81         }
  82 }
  83
  84 //**********************************************************************************************************************
  85 GetSharedOTUCommand::GetSharedOTUCommand(){
  86         try {
  87                 abort = true; calledHelp = true;
  88                 setParameters();
  89                 vector<string> tempOutNames;
  90                 outputTypes["fasta"] = tempOutNames;
  91                 outputTypes["accnos"] = tempOutNames;
  92                 outputTypes["sharedseqs"] = tempOutNames;
  93         }
  94         catch(exception& e) {
  95                 m->errorOut(e, "GetSharedOTUCommand", "GetSharedOTUCommand");
  96                 exit(1);
  97         }
  98 }
  99 //**********************************************************************************************************************
 100 GetSharedOTUCommand::GetSharedOTUCommand(string option)  {
 101         try {
 102
 103                 abort = false; calledHelp = false;
 104                 unique = true;
 105                 allLines = 1;
 106
 107                 //allow user to run help
 108                 if(option == "help") { help(); abort = true; calledHelp = true; }
 109                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 110
 111                 else {
 112                         vector<string> myArray = setParameters();
 113
 114                         OptionParser parser(option);
 115                         map<string,string> parameters = parser.getParameters();
 116
 117                         ValidParameters validParameter;
 118                         map<string,string>::iterator it;
 119
 120                         //check to make sure all parameters are valid for command
 121                         for (it = parameters.begin(); it != parameters.end(); it++) {
 122                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 123                         }
 124
 125                         //initialize outputTypes
 126                         vector<string> tempOutNames;
 127                         outputTypes["fasta"] = tempOutNames;
 128                         outputTypes["accnos"] = tempOutNames;
 129                         outputTypes["sharedseqs"] = tempOutNames;
 130
 131                         //if the user changes the output directory command factory will send this info to us in the output parameter
 132                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 133
 134                         //if the user changes the input directory command factory will send this info to us in the output parameter
 135                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 136                         if (inputDir == "not found"){   inputDir = "";          }
 137                         else {
 138                                 string path;
 139                                 it = parameters.find("fasta");
 140                                 //user has given a template file
 141                                 if(it != parameters.end()){
 142                                         path = m->hasPath(it->second);
 143                                         //if the user has not given a path then, add inputdir. else leave path alone.
 144                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 145                                 }
 146
 147                                 it = parameters.find("list");
 148                                 //user has given a template file
 149                                 if(it != parameters.end()){
 150                                         path = m->hasPath(it->second);
 151                                         //if the user has not given a path then, add inputdir. else leave path alone.
 152                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 153                                 }
 154
 155                                 it = parameters.find("group");
 156                                 //user has given a template file
 157                                 if(it != parameters.end()){
 158                                         path = m->hasPath(it->second);
 159                                         //if the user has not given a path then, add inputdir. else leave path alone.
 160                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 161                                 }
 162                         }
 163
 164
 165                         //check for required parameters
 166                         listfile = validParameter.validFile(parameters, "list", true);
 167                         if (listfile == "not open") { abort = true; }
 168                         else if (listfile == "not found") {
 169                                 listfile = m->getListFile();
 170                                 if (listfile != "") { format = "list"; m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 171                                 else {
 172                                         m->mothurOut("No valid current list file. You must provide a list file."); m->mothurOutEndLine();
 173                                         abort = true;
 174                                 }
 175                         }else {  format = "list";       m->setListFile(listfile); }
 176
 177                         groupfile = validParameter.validFile(parameters, "group", true);
 178                         if (groupfile == "not open") { abort = true; }
 179                         else if (groupfile == "not found") {
 180                                 groupfile = m->getGroupFile();
 181                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 182                                 else {
 183                                         m->mothurOut("No valid current group file. You must provide a group file."); m->mothurOutEndLine();
 184                                         abort = true;
 185                                 }
 186                         }else { m->setGroupFile(groupfile); }
 187
 188                         if ((listfile == "") || (groupfile == "")) { m->mothurOut("The list and group parameters are required."); m->mothurOutEndLine(); abort = true; }
 189
 190                         //check for optional parameter and set defaults
 191                         // ...at some point should added some additional type checking...
 192                         label = validParameter.validFile(parameters, "label", false);
 193                         if (label == "not found") { label = ""; }
 194                         else {
 195                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 196                                 else { allLines = 1;  }
 197                         }
 198
 199                         output = validParameter.validFile(parameters, "output", false);
 200                         if (output == "not found") { output = ""; }
 201                         else if (output == "default") { output = ""; }
 202
 203                         groups = validParameter.validFile(parameters, "unique", false);
 204                         if (groups == "not found") { groups = ""; }
 205                         else {
 206                                 userGroups = "unique." + groups;
 207                                 m->splitAtDash(groups, Groups);
 208                                 m->setGroups(Groups);
 209
 210                         }
 211
 212                         groups = validParameter.validFile(parameters, "shared", false);
 213                         if (groups == "not found") { groups = "";  }
 214                         else {
 215                                 userGroups = groups;
 216                                 m->splitAtDash(groups, Groups);
 217                                 m->setGroups(Groups);
 218                                 unique = false;
 219                         }
 220
 221                         fastafile = validParameter.validFile(parameters, "fasta", true);
 222                         if (fastafile == "not open") { abort = true; }
 223                         else if (fastafile == "not found") {  fastafile = "";  }
 224                         else { m->setFastaFile(fastafile); }
 225                 }
 226
 227         }
 228         catch(exception& e) {
 229                 m->errorOut(e, "GetSharedOTUCommand", "GetSharedOTUCommand");
 230                 exit(1);
 231         }
 232 }
 233 //**********************************************************************************************************************
 234
 235 int GetSharedOTUCommand::execute(){
 236         try {
 237
 238                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 239
 240                 groupMap = new GroupMap(groupfile);
 241                 int error = groupMap->readMap();
 242                 if (error == 1) { delete groupMap; return 0; }
 243
 244                 if (m->control_pressed) { delete groupMap; return 0; }
 245
 246                 if (Groups.size() == 0) {
 247                         Groups = groupMap->getNamesOfGroups();
 248
 249                         //make string for outputfile name
 250                         userGroups = "unique.";
 251                         for(int i = 0; i < Groups.size(); i++) {  userGroups += Groups[i] + "-";  }
 252                         userGroups = userGroups.substr(0, userGroups.length()-1);
 253                 }else{
 254                         //sanity check for group names
 255                         SharedUtil util;
 256                         vector<string> namesOfGroups = groupMap->getNamesOfGroups();
 257                         util.setGroups(Groups, namesOfGroups);
 258                         groupMap->setNamesOfGroups(namesOfGroups);
 259                 }
 260
 261                 //put groups in map to find easier
 262                 for(int i = 0; i < Groups.size(); i++) {
 263                         groupFinder[Groups[i]] = Groups[i];
 264                 }
 265
 266                 if (fastafile != "") {
 267                         ifstream inFasta;
 268                         m->openInputFile(fastafile, inFasta);
 269
 270                         while(!inFasta.eof()) {
 271                                 if (m->control_pressed) { outputTypes.clear(); inFasta.close(); delete groupMap; return 0; }
 272
 273                                 Sequence seq(inFasta); m->gobble(inFasta);
 274                                 if (seq.getName() != "") {  seqs.push_back(seq);   }
 275                         }
 276                         inFasta.close();
 277                 }
 278
 279                 ListVector* lastlist = NULL;
 280                 string lastLabel = "";
 281
 282                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 283                 set<string> processedLabels;
 284                 set<string> userLabels = labels;
 285
 286                 ifstream in;
 287                 m->openInputFile(listfile, in);
 288
 289                 //as long as you are not at the end of the file or done wih the lines you want
 290                 while((!in.eof()) && ((allLines == 1) || (userLabels.size() != 0))) {
 291
 292                         if (m->control_pressed) {
 293                                 if (lastlist != NULL) {         delete lastlist;        }
 294                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); }  outputTypes.clear();
 295                                 delete groupMap; return 0;
 296                         }
 297
 298                         list = new ListVector(in);
 299
 300                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 301                                 m->mothurOut(list->getLabel());
 302                                 process(list);
 303
 304                                 processedLabels.insert(list->getLabel());
 305                                 userLabels.erase(list->getLabel());
 306                         }
 307
 308                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 309                                         string saveLabel = list->getLabel();
 310
 311                                         m->mothurOut(lastlist->getLabel());
 312                                         process(lastlist);
 313
 314                                         processedLabels.insert(lastlist->getLabel());
 315                                         userLabels.erase(lastlist->getLabel());
 316
 317                                         //restore real lastlabel to save below
 318                                         list->setLabel(saveLabel);
 319                         }
 320
 321                         lastLabel = list->getLabel();
 322
 323                         if (lastlist != NULL) {         delete lastlist;        }
 324                         lastlist = list;
 325                 }
 326
 327                 in.close();
 328
 329                 //output error messages about any remaining user labels
 330                 set<string>::iterator it;
 331                 bool needToRun = false;
 332                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 333                         m->mothurOut("Your file does not include the label " + *it);
 334                         if (processedLabels.count(lastLabel) != 1) {
 335                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 336                                 needToRun = true;
 337                         }else {
 338                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 339                         }
 340                 }
 341
 342                 //run last label if you need to
 343                 if (needToRun == true)  {
 344                                 m->mothurOut(lastlist->getLabel());
 345                                 process(lastlist);
 346
 347                                 processedLabels.insert(lastlist->getLabel());
 348                                 userLabels.erase(lastlist->getLabel());
 349                 }
 350
 351
 352                 //reset groups parameter
 353                 m->clearGroups();
 354
 355                 if (lastlist != NULL) {         delete lastlist;        }
 356
 357                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]); }  delete groupMap; return 0; }
 358
 359                 //set fasta file as new current fastafile
 360                 string current = "";
 361                 itTypes = outputTypes.find("fasta");
 362                 if (itTypes != outputTypes.end()) {
 363                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 364                 }
 365
 366                 if (output == "accnos") {
 367                         itTypes = outputTypes.find("accnos");
 368                         if (itTypes != outputTypes.end()) {
 369                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 370                         }
 371                 }
 372
 373                 m->mothurOutEndLine();
 374                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 375                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 376                 m->mothurOutEndLine();
 377
 378
 379                 return 0;
 380         }
 381
 382         catch(exception& e) {
 383                 m->errorOut(e, "GetSharedOTUCommand", "execute");
 384                 exit(1);
 385         }
 386 }
 387 /***********************************************************/
 388 int GetSharedOTUCommand::process(ListVector* shared) {
 389         try {
 390
 391                 map<string, string> fastaMap;
 392
 393                 ofstream outNames;
 394                 string outputFileNames;
 395
 396                 if (outputDir == "") { outputDir += m->hasPath(listfile); }
 397                 if (output != "accnos") {
 398                         outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + "." + getOutputFileNameTag("sharedseqs");
 399                 }else {
 400                         outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + "." + getOutputFileNameTag("accnos");
 401                 }
 402                 m->openOutputFile(outputFileNames, outNames);
 403
 404                 bool wroteSomething = false;
 405                 int num = 0;
 406
 407                 //go through each bin, find out if shared
 408                 for (int i = 0; i < shared->getNumBins(); i++) {
 409                         if (m->control_pressed) { outNames.close(); m->mothurRemove(outputFileNames); return 0; }
 410
 411                         bool uniqueOTU = true;
 412
 413                         map<string, int> atLeastOne;
 414                         for (int f = 0; f < Groups.size(); f++) {
 415                                 atLeastOne[Groups[f]] = 0;
 416                         }
 417
 418                         vector<string> namesOfSeqsInThisBin;
 419
 420                         string names = shared->get(i);
 421             vector<string> binNames;
 422             m->splitAtComma(names, binNames);
 423                         for(int j = 0; j < binNames.size(); j++) {
 424                                 string name = binNames[j];
 425
 426                                 //find group
 427                                 string seqGroup = groupMap->getGroup(name);
 428                                 if (output != "accnos") {
 429                                         namesOfSeqsInThisBin.push_back((name + "|" + seqGroup + "|" + toString(i+1)));
 430                                 }else {  namesOfSeqsInThisBin.push_back(name);  }
 431
 432                                 if (seqGroup == "not found") { m->mothurOut(name + " is not in your groupfile. Please correct."); m->mothurOutEndLine(); exit(1);  }
 433
 434                                 //is this seq in one of hte groups we care about
 435                                 it = groupFinder.find(seqGroup);
 436                                 if (it == groupFinder.end()) {  uniqueOTU = false;  } //you have a sequence from a group you don't want
 437                                 else {  atLeastOne[seqGroup]++;  }
 438                         }
 439
 440                         //make sure you have at least one seq from each group you want
 441                         bool sharedByAll = true;
 442                         map<string, int>::iterator it2;
 443                         for (it2 = atLeastOne.begin(); it2 != atLeastOne.end(); it2++) {
 444                                 if (it2->second == 0) {  sharedByAll = false;   }
 445                         }
 446
 447                         //if the user wants unique bins and this is unique then print
 448                         //or this the user wants shared bins and this bin is shared then print
 449                         if ((unique && uniqueOTU && sharedByAll) || (!unique && sharedByAll)) {
 450
 451                                 wroteSomething = true;
 452                                 num++;
 453
 454                                 //output list of names
 455                                 for (int j = 0; j < namesOfSeqsInThisBin.size(); j++) {
 456                                         outNames << namesOfSeqsInThisBin[j] << endl;
 457
 458                                         if (fastafile != "") {
 459                                                 if (output != "accnos") {
 460                                                         string seqName = namesOfSeqsInThisBin[j].substr(0,namesOfSeqsInThisBin[j].find_last_of('|'));
 461                                                         seqName = seqName.substr(0,seqName.find_last_of('|'));
 462                                                         fastaMap[seqName] = namesOfSeqsInThisBin[j];  //fastaMap needs to contain just the seq name for output later
 463                                                 }else {
 464                                                         fastaMap[namesOfSeqsInThisBin[j]] = namesOfSeqsInThisBin[j];
 465                                                 }
 466                                         }
 467                                 }
 468                         }
 469                 }
 470
 471                 outNames.close();
 472
 473                 if (!wroteSomething) {
 474                         m->mothurRemove(outputFileNames);
 475                         string outputString = "\t" + toString(num) + " - No otus shared by groups";
 476
 477                         string groupString = "";
 478                         for (int h = 0; h < Groups.size(); h++) {
 479                                 groupString += "  " + Groups[h];
 480                         }
 481
 482                         outputString += groupString + ".";
 483                         m->mothurOut(outputString); m->mothurOutEndLine();
 484                 }else {
 485                         m->mothurOut("\t" + toString(num)); m->mothurOutEndLine();
 486                         outputNames.push_back(outputFileNames);
 487                         if (output != "accnos") { outputTypes["sharedseqs"].push_back(outputFileNames); }
 488                         else { outputTypes["accnos"].push_back(outputFileNames); }
 489                 }
 490
 491                 //if fasta file provided output new fasta file
 492                 if ((fastafile != "") && wroteSomething) {
 493                         if (outputDir == "") { outputDir += m->hasPath(fastafile); }
 494                         string outputFileFasta = outputDir + m->getRootName(m->getSimpleName(fastafile)) + shared->getLabel() + userGroups + "." + getOutputFileNameTag("fasta");
 495                         ofstream outFasta;
 496                         m->openOutputFile(outputFileFasta, outFasta);
 497                         outputNames.push_back(outputFileFasta); outputTypes["fasta"].push_back(outputFileFasta);
 498
 499                         for (int k = 0; k < seqs.size(); k++) {
 500                                 if (m->control_pressed) { outFasta.close(); return 0; }
 501
 502                                 //if this is a sequence we want, output it
 503                                 it = fastaMap.find(seqs[k].getName());
 504                                 if (it != fastaMap.end()) {
 505
 506                                         if (output != "accnos") {
 507                                                 outFasta << ">" << it->second << endl;
 508                                         }else {
 509                                                 outFasta << ">" << it->first << endl;
 510                                         }
 511
 512                                         outFasta << seqs[k].getAligned() << endl;
 513                                 }
 514                         }
 515
 516                         outFasta.close();
 517                 }
 518
 519                 return 0;
 520
 521         }
 522         catch(exception& e) {
 523                 m->errorOut(e, "GetSharedOTUCommand", "process");
 524                 exit(1);
 525         }
 526 }
 527
 528 //**********************************************************************************************************************