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