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