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