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