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