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
 139                         groupfile = validParameter.validFile(parameters, "group", true);
 140                         if (groupfile == "not open") { abort = true; }
 141                         else if (groupfile == "not found") {
 142                                 groupfile = m->getGroupFile();
 143                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 144                                 else {  m->mothurOut("You have no current group file and the group parameter is required."); m->mothurOutEndLine(); abort = true; }
 145                         }
 146
 147                         listfile = validParameter.validFile(parameters, "list", true);
 148                         if (listfile == "not open") { abort = true; }
 149                         else if (listfile == "not found") {
 150                                 listfile = m->getListFile();
 151                                 if (listfile != "") { m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 152                                 else {  m->mothurOut("You have no current list file and the list parameter is required."); m->mothurOutEndLine(); abort = true; }
 153                         }
 154
 155                         groups = validParameter.validFile(parameters, "groups", false);
 156                         if (groups == "not found") { groups = ""; }
 157                         else {
 158                                 m->splitAtDash(groups, Groups);
 159                         }
 160
 161                         label = validParameter.validFile(parameters, "label", false);
 162                         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=""; }
 163
 164                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 165                 }
 166
 167         }
 168         catch(exception& e) {
 169                 m->errorOut(e, "RemoveOtusCommand", "RemoveOtusCommand");
 170                 exit(1);
 171         }
 172 }
 173 //**********************************************************************************************************************
 174
 175 int RemoveOtusCommand::execute(){
 176         try {
 177
 178                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 179
 180                 groupMap = new GroupMap(groupfile);
 181                 groupMap->readMap();
 182
 183                 //get groups you want to remove
 184                 if (accnosfile != "") { readAccnos(); }
 185
 186                 //make sure groups are valid
 187                 //takes care of user setting groupNames that are invalid or setting groups=all
 188                 SharedUtil* util = new SharedUtil();
 189                 util->setGroups(Groups, groupMap->namesOfGroups);
 190                 delete util;
 191
 192                 if (m->control_pressed) { delete groupMap; return 0; }
 193
 194                 //read through the list file keeping any otus that contain any sequence from the groups selected
 195                 readListGroup();
 196
 197                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 198
 199                 if (outputNames.size() != 0) {
 200                         m->mothurOutEndLine();
 201                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 202                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 203                         m->mothurOutEndLine();
 204
 205                         //set fasta file as new current fastafile
 206                         string current = "";
 207                         itTypes = outputTypes.find("group");
 208                         if (itTypes != outputTypes.end()) {
 209                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 210                         }
 211
 212                         itTypes = outputTypes.find("list");
 213                         if (itTypes != outputTypes.end()) {
 214                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 215                         }
 216                 }
 217
 218                 return 0;
 219         }
 220
 221         catch(exception& e) {
 222                 m->errorOut(e, "RemoveOtusCommand", "execute");
 223                 exit(1);
 224         }
 225 }
 226 //**********************************************************************************************************************
 227 int RemoveOtusCommand::readListGroup(){
 228         try {
 229                 string thisOutputDir = outputDir;
 230                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 231                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick." + label +  m->getExtension(listfile);
 232
 233                 ofstream out;
 234                 m->openOutputFile(outputFileName, out);
 235
 236                 string GroupOutputDir = outputDir;
 237                 if (outputDir == "") {  GroupOutputDir += m->hasPath(groupfile);  }
 238                 string outputGroupFileName = GroupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick." + label  + m->getExtension(groupfile);
 239
 240                 ofstream outGroup;
 241                 m->openOutputFile(outputGroupFileName, outGroup);
 242
 243                 InputData* input = new InputData(listfile, "list");
 244                 ListVector* list = input->getListVector();
 245                 string lastLabel = list->getLabel();
 246
 247                 //using first label seen if none is provided
 248                 if (label == "") { label = lastLabel; }
 249
 250                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 251                 set<string> labels; labels.insert(label);
 252                 set<string> processedLabels;
 253                 set<string> userLabels = labels;
 254
 255                 bool wroteSomething = false;
 256
 257                 //as long as you are not at the end of the file or done wih the lines you want
 258                 while((list != NULL) && (userLabels.size() != 0)) {
 259
 260                         if (m->control_pressed) {  delete list; delete input; out.close();  outGroup.close(); remove(outputFileName.c_str());  remove(outputGroupFileName.c_str());return 0;  }
 261
 262                         if(labels.count(list->getLabel()) == 1){
 263                                 processList(list, groupMap, out, outGroup, wroteSomething);
 264
 265                                 processedLabels.insert(list->getLabel());
 266                                 userLabels.erase(list->getLabel());
 267                         }
 268
 269                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 270                                 string saveLabel = list->getLabel();
 271
 272                                 delete list;
 273
 274                                 list = input->getListVector(lastLabel);
 275
 276                                 processList(list, groupMap, out, outGroup, wroteSomething);
 277
 278                                 processedLabels.insert(list->getLabel());
 279                                 userLabels.erase(list->getLabel());
 280
 281                                 //restore real lastlabel to save below
 282                                 list->setLabel(saveLabel);
 283                         }
 284
 285                         lastLabel = list->getLabel();
 286
 287                         delete list; list = NULL;
 288
 289                         //get next line to process
 290                         list = input->getListVector();
 291                 }
 292
 293
 294                 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;  }
 295
 296                 //output error messages about any remaining user labels
 297                 set<string>::iterator it;
 298                 bool needToRun = false;
 299                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 300                         m->mothurOut("Your file does not include the label " + *it);
 301                         if (processedLabels.count(lastLabel) != 1) {
 302                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 303                                 needToRun = true;
 304                         }else {
 305                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 306                         }
 307                 }
 308
 309                 //run last label if you need to
 310                 if (needToRun == true)  {
 311                         if (list != NULL) { delete list; }
 312
 313                         list = input->getListVector(lastLabel);
 314
 315                         processList(list, groupMap, out, outGroup, wroteSomething);
 316
 317                         delete list; list = NULL;
 318                 }
 319
 320                 out.close();
 321                 outGroup.close();
 322
 323                 if (wroteSomething == false) {  m->mothurOut("At distance " + label + " your file ONLY contains otus containing sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 324                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 325                 outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName);
 326
 327                 return 0;
 328
 329         }
 330         catch(exception& e) {
 331                 m->errorOut(e, "RemoveOtusCommand", "readList");
 332                 exit(1);
 333         }
 334 }
 335 //**********************************************************************************************************************
 336 int RemoveOtusCommand::processList(ListVector*& list, GroupMap*& groupMap, ofstream& out, ofstream& outGroup, bool& wroteSomething){
 337         try {
 338
 339                 //make a new list vector
 340                 ListVector newList;
 341                 newList.setLabel(list->getLabel());
 342
 343                 int numOtus = 0;
 344                 //for each bin
 345                 for (int i = 0; i < list->getNumBins(); i++) {
 346                         if (m->control_pressed) { return 0; }
 347
 348                         //parse out names that are in accnos file
 349                         string binnames = list->get(i);
 350
 351                         bool removeBin = false;
 352                         string groupFileOutput = "";
 353
 354                         //parse names
 355                         string individual = "";
 356                         int length = binnames.length();
 357                         for(int j=0;j<length;j++){
 358                                 if(binnames[j] == ','){
 359                                         string group = groupMap->getGroup(individual);
 360                                         if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 361
 362                                         if (m->inUsersGroups(group, Groups)) {  removeBin = true; break; }
 363                                         groupFileOutput += individual + "\t" + group + "\n";
 364                                         individual = "";
 365
 366                                 }
 367                                 else{  individual += binnames[j];  }
 368                         }
 369
 370                         if (!removeBin) {
 371                                 //get last name
 372                                 string group = groupMap->getGroup(individual);
 373                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 374
 375                                 if (m->inUsersGroups(group, Groups)) {  removeBin = true; }
 376                                 groupFileOutput += individual + "\t" + group + "\n";
 377
 378                                 if (!removeBin) {
 379                                         //if there are no sequences from the groups we want to remove in this bin add to new list, output to groupfile
 380                                         newList.push_back(binnames);
 381                                         outGroup << groupFileOutput;
 382                                 }else {
 383                                         numOtus++;
 384                                 }
 385                         }else {
 386                                 numOtus++;
 387                         }
 388
 389                 }
 390
 391                 //print new listvector
 392                 if (newList.getNumBins() != 0) {
 393                         wroteSomething = true;
 394                         newList.print(out);
 395                 }
 396
 397                 m->mothurOut(newList.getLabel() + " - removed " + toString(numOtus) + " of the " + toString(list->getNumBins()) + " OTUs."); m->mothurOutEndLine();
 398
 399                 return 0;
 400
 401         }
 402         catch(exception& e) {
 403                 m->errorOut(e, "RemoveOtusCommand", "processList");
 404                 exit(1);
 405         }
 406 }
 407 //**********************************************************************************************************************
 408 void RemoveOtusCommand::readAccnos(){
 409         try {
 410                 Groups.clear();
 411
 412                 ifstream in;
 413                 m->openInputFile(accnosfile, in);
 414                 string name;
 415
 416                 while(!in.eof()){
 417                         in >> name;
 418
 419                         Groups.push_back(name);
 420
 421                         m->gobble(in);
 422                 }
 423                 in.close();
 424
 425         }
 426         catch(exception& e) {
 427                 m->errorOut(e, "RemoveOtusCommand", "readAccnos");
 428                 exit(1);
 429         }
 430 }
 431 //**********************************************************************************************************************
 432
 433
 434