removeotuscommand.cpp

   1 /*
   2  *  removeotuscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/12/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "removeotuscommand.h"
  11 #include "inputdata.h"
  12 #include "sharedutilities.h"
  13
  14
  15 //**********************************************************************************************************************
  16 vector<string> RemoveOtusCommand::getValidParameters(){
  17         try {
  18                 string Array[] =  { "group", "accnos","label", "groups","list","outputdir","inputdir" };
  19                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  20                 return myArray;
  21         }
  22         catch(exception& e) {
  23                 m->errorOut(e, "RemoveOtusCommand", "getValidParameters");
  24                 exit(1);
  25         }
  26 }
  27 //**********************************************************************************************************************
  28 RemoveOtusCommand::RemoveOtusCommand(){
  29         try {
  30                 abort = true; calledHelp = true;
  31                 vector<string> tempOutNames;
  32                 outputTypes["group"] = tempOutNames;
  33                 outputTypes["list"] = tempOutNames;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "RemoveOtusCommand", "RemoveOtusCommand");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 vector<string> RemoveOtusCommand::getRequiredParameters(){
  42         try {
  43                 string Array[] =  {"group","label", "list"};
  44                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  45                 return myArray;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "RemoveOtusCommand", "getRequiredParameters");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 vector<string> RemoveOtusCommand::getRequiredFiles(){
  54         try {
  55                 vector<string> myArray;
  56                 return myArray;
  57         }
  58         catch(exception& e) {
  59                 m->errorOut(e, "RemoveOtusCommand", "getRequiredFiles");
  60                 exit(1);
  61         }
  62 }
  63 //**********************************************************************************************************************
  64 RemoveOtusCommand::RemoveOtusCommand(string option)  {
  65         try {
  66                 abort = false; calledHelp = false;
  67
  68                 //allow user to run help
  69                 if(option == "help") { help(); abort = true; calledHelp = true; }
  70
  71                 else {
  72                         //valid paramters for this command
  73                         string Array[] =  { "group", "accnos","label", "groups", "list","outputdir","inputdir" };
  74                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  75
  76                         OptionParser parser(option);
  77                         map<string,string> parameters = parser.getParameters();
  78
  79                         ValidParameters validParameter;
  80                         map<string,string>::iterator it;
  81
  82                         //check to make sure all parameters are valid for command
  83                         for (it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86
  87                         //initialize outputTypes
  88                         vector<string> tempOutNames;
  89                         outputTypes["group"] = tempOutNames;
  90                         outputTypes["list"] = tempOutNames;
  91
  92
  93                         //if the user changes the output directory command factory will send this info to us in the output parameter
  94                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  95
  96                         //if the user changes the input directory command factory will send this info to us in the output parameter
  97                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  98                         if (inputDir == "not found"){   inputDir = "";          }
  99                         else {
 100                                 string path;
 101                                 it = parameters.find("accnos");
 102                                 //user has given a template file
 103                                 if(it != parameters.end()){
 104                                         path = m->hasPath(it->second);
 105                                         //if the user has not given a path then, add inputdir. else leave path alone.
 106                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 107                                 }
 108
 109                                 it = parameters.find("list");
 110                                 //user has given a template file
 111                                 if(it != parameters.end()){
 112                                         path = m->hasPath(it->second);
 113                                         //if the user has not given a path then, add inputdir. else leave path alone.
 114                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 115                                 }
 116
 117                                 it = parameters.find("group");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 123                                 }
 124                         }
 125
 126
 127                         //check for required parameters
 128                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 129                         if (accnosfile == "not open") { abort = true; }
 130                         else if (accnosfile == "not found") {  accnosfile = ""; }
 131
 132                         groupfile = validParameter.validFile(parameters, "group", true);
 133                         if (groupfile == "not open") { abort = true; }
 134                         else if (groupfile == "not found") {  groupfile = "";  m->mothurOut("You must provide a group file."); m->mothurOutEndLine(); abort = true; }
 135
 136                         listfile = validParameter.validFile(parameters, "list", true);
 137                         if (listfile == "not open") { abort = true; }
 138                         else if (listfile == "not found") {  listfile = ""; m->mothurOut("You must provide a list file."); m->mothurOutEndLine(); abort = true; }
 139
 140                         groups = validParameter.validFile(parameters, "groups", false);
 141                         if (groups == "not found") { groups = ""; }
 142                         else {
 143                                 m->splitAtDash(groups, Groups);
 144                         }
 145
 146                         label = validParameter.validFile(parameters, "label", false);
 147                         if (label == "not found") { label = ""; m->mothurOut("You must provide a label to process."); m->mothurOutEndLine(); abort = true; }
 148
 149                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 150                 }
 151
 152         }
 153         catch(exception& e) {
 154                 m->errorOut(e, "RemoveOtusCommand", "RemoveOtusCommand");
 155                 exit(1);
 156         }
 157 }
 158 //**********************************************************************************************************************
 159
 160 void RemoveOtusCommand::help(){
 161         try {
 162                 m->mothurOut("The remove.otus command removes otus containing sequences from a specfic group or set of groups.\n");
 163                 m->mothurOut("It outputs a new list file containing the otus containing sequences NOT from in the those specified groups.\n");
 164                 m->mothurOut("The remove.otus command parameters are accnos, group, list, label and groups. The group, list and label parameters are required.\n");
 165                 m->mothurOut("You must also provide an accnos containing the list of groups to get or set the groups parameter to the groups you wish to select.\n");
 166                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like.  You can separate group names with dashes.\n");
 167                 m->mothurOut("The label parameter allows you to specify which distance you want to process.\n");
 168                 m->mothurOut("The remove.otus command should be in the following format: remove.otus(accnos=yourAccnos, list=yourListFile, group=yourGroupFile, label=yourLabel).\n");
 169                 m->mothurOut("Example remove.otus(accnos=amazon.accnos, list=amazon.fn.list, group=amazon.groups, label=0.03).\n");
 170                 m->mothurOut("or remove.otus(groups=pasture, list=amazon.fn.list, amazon.groups, label=0.03).\n");
 171                 m->mothurOut("Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListFile).\n\n");
 172         }
 173         catch(exception& e) {
 174                 m->errorOut(e, "RemoveOtusCommand", "help");
 175                 exit(1);
 176         }
 177 }
 178
 179 //**********************************************************************************************************************
 180
 181 int RemoveOtusCommand::execute(){
 182         try {
 183
 184                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 185
 186                 groupMap = new GroupMap(groupfile);
 187                 groupMap->readMap();
 188
 189                 //get groups you want to remove
 190                 if (accnosfile != "") { readAccnos(); }
 191
 192                 //make sure groups are valid
 193                 //takes care of user setting groupNames that are invalid or setting groups=all
 194                 SharedUtil* util = new SharedUtil();
 195                 util->setGroups(Groups, groupMap->namesOfGroups);
 196                 delete util;
 197
 198                 if (m->control_pressed) { delete groupMap; return 0; }
 199
 200                 //read through the list file keeping any otus that contain any sequence from the groups selected
 201                 readListGroup();
 202
 203                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 204
 205                 if (outputNames.size() != 0) {
 206                         m->mothurOutEndLine();
 207                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 208                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 209                         m->mothurOutEndLine();
 210
 211                         //set fasta file as new current fastafile
 212                         string current = "";
 213                         itTypes = outputTypes.find("group");
 214                         if (itTypes != outputTypes.end()) {
 215                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 216                         }
 217
 218                         itTypes = outputTypes.find("list");
 219                         if (itTypes != outputTypes.end()) {
 220                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 221                         }
 222                 }
 223
 224                 return 0;
 225         }
 226
 227         catch(exception& e) {
 228                 m->errorOut(e, "RemoveOtusCommand", "execute");
 229                 exit(1);
 230         }
 231 }
 232 //**********************************************************************************************************************
 233 int RemoveOtusCommand::readListGroup(){
 234         try {
 235                 string thisOutputDir = outputDir;
 236                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 237                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick." + label +  m->getExtension(listfile);
 238
 239                 ofstream out;
 240                 m->openOutputFile(outputFileName, out);
 241
 242                 string GroupOutputDir = outputDir;
 243                 if (outputDir == "") {  GroupOutputDir += m->hasPath(groupfile);  }
 244                 string outputGroupFileName = GroupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick." + label  + m->getExtension(groupfile);
 245
 246                 ofstream outGroup;
 247                 m->openOutputFile(outputGroupFileName, outGroup);
 248
 249                 InputData* input = new InputData(listfile, "list");
 250                 ListVector* list = input->getListVector();
 251                 string lastLabel = list->getLabel();
 252
 253                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 254                 set<string> labels; labels.insert(label);
 255                 set<string> processedLabels;
 256                 set<string> userLabels = labels;
 257
 258                 bool wroteSomething = false;
 259
 260                 //as long as you are not at the end of the file or done wih the lines you want
 261                 while((list != NULL) && (userLabels.size() != 0)) {
 262
 263                         if (m->control_pressed) {  delete list; delete input; out.close();  outGroup.close(); remove(outputFileName.c_str());  remove(outputGroupFileName.c_str());return 0;  }
 264
 265                         if(labels.count(list->getLabel()) == 1){
 266                                 processList(list, groupMap, out, outGroup, wroteSomething);
 267
 268                                 processedLabels.insert(list->getLabel());
 269                                 userLabels.erase(list->getLabel());
 270                         }
 271
 272                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 273                                 string saveLabel = list->getLabel();
 274
 275                                 delete list;
 276
 277                                 list = input->getListVector(lastLabel);
 278
 279                                 processList(list, groupMap, out, outGroup, wroteSomething);
 280
 281                                 processedLabels.insert(list->getLabel());
 282                                 userLabels.erase(list->getLabel());
 283
 284                                 //restore real lastlabel to save below
 285                                 list->setLabel(saveLabel);
 286                         }
 287
 288                         lastLabel = list->getLabel();
 289
 290                         delete list; list = NULL;
 291
 292                         //get next line to process
 293                         list = input->getListVector();
 294                 }
 295
 296
 297                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); outGroup.close(); remove(outputFileName.c_str());  remove(outputGroupFileName.c_str()); return 0;  }
 298
 299                 //output error messages about any remaining user labels
 300                 set<string>::iterator it;
 301                 bool needToRun = false;
 302                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 303                         m->mothurOut("Your file does not include the label " + *it);
 304                         if (processedLabels.count(lastLabel) != 1) {
 305                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 306                                 needToRun = true;
 307                         }else {
 308                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 309                         }
 310                 }
 311
 312                 //run last label if you need to
 313                 if (needToRun == true)  {
 314                         if (list != NULL) { delete list; }
 315
 316                         list = input->getListVector(lastLabel);
 317
 318                         processList(list, groupMap, out, outGroup, wroteSomething);
 319
 320                         delete list; list = NULL;
 321                 }
 322
 323                 out.close();
 324                 outGroup.close();
 325
 326                 if (wroteSomething == false) {  m->mothurOut("At distance " + label + " your file ONLY contains otus containing sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 327                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 328                 outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName);
 329
 330                 return 0;
 331
 332         }
 333         catch(exception& e) {
 334                 m->errorOut(e, "RemoveOtusCommand", "readList");
 335                 exit(1);
 336         }
 337 }
 338 //**********************************************************************************************************************
 339 int RemoveOtusCommand::processList(ListVector*& list, GroupMap*& groupMap, ofstream& out, ofstream& outGroup, bool& wroteSomething){
 340         try {
 341
 342                 //make a new list vector
 343                 ListVector newList;
 344                 newList.setLabel(list->getLabel());
 345
 346                 int numOtus = 0;
 347                 //for each bin
 348                 for (int i = 0; i < list->getNumBins(); i++) {
 349                         if (m->control_pressed) { return 0; }
 350
 351                         //parse out names that are in accnos file
 352                         string binnames = list->get(i);
 353
 354                         bool removeBin = false;
 355                         string groupFileOutput = "";
 356
 357                         //parse names
 358                         string individual = "";
 359                         int length = binnames.length();
 360                         for(int j=0;j<length;j++){
 361                                 if(binnames[j] == ','){
 362                                         string group = groupMap->getGroup(individual);
 363                                         if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 364
 365                                         if (m->inUsersGroups(group, Groups)) {  removeBin = true; break; }
 366                                         groupFileOutput += individual + "\t" + group + "\n";
 367                                         individual = "";
 368
 369                                 }
 370                                 else{  individual += binnames[j];  }
 371                         }
 372
 373                         if (!removeBin) {
 374                                 //get last name
 375                                 string group = groupMap->getGroup(individual);
 376                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 377
 378                                 if (m->inUsersGroups(group, Groups)) {  removeBin = true; }
 379                                 groupFileOutput += individual + "\t" + group + "\n";
 380
 381                                 if (!removeBin) {
 382                                         //if there are no sequences from the groups we want to remove in this bin add to new list, output to groupfile
 383                                         newList.push_back(binnames);
 384                                         outGroup << groupFileOutput;
 385                                 }else {
 386                                         numOtus++;
 387                                 }
 388                         }else {
 389                                 numOtus++;
 390                         }
 391
 392                 }
 393
 394                 //print new listvector
 395                 if (newList.getNumBins() != 0) {
 396                         wroteSomething = true;
 397                         newList.print(out);
 398                 }
 399
 400                 m->mothurOut(newList.getLabel() + " - removed " + toString(numOtus) + " of the " + toString(list->getNumBins()) + " OTUs."); m->mothurOutEndLine();
 401
 402                 return 0;
 403
 404         }
 405         catch(exception& e) {
 406                 m->errorOut(e, "RemoveOtusCommand", "processList");
 407                 exit(1);
 408         }
 409 }
 410 //**********************************************************************************************************************
 411 void RemoveOtusCommand::readAccnos(){
 412         try {
 413                 Groups.clear();
 414
 415                 ifstream in;
 416                 m->openInputFile(accnosfile, in);
 417                 string name;
 418
 419                 while(!in.eof()){
 420                         in >> name;
 421
 422                         Groups.push_back(name);
 423
 424                         m->gobble(in);
 425                 }
 426                 in.close();
 427
 428         }
 429         catch(exception& e) {
 430                 m->errorOut(e, "RemoveOtusCommand", "readAccnos");
 431                 exit(1);
 432         }
 433 }
 434 //**********************************************************************************************************************
 435
 436
 437