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