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