getotuscommand.cpp

   1 /*
   2  *  getotuscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/10/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "getotuscommand.h"
  11 #include "inputdata.h"
  12 #include "sharedutilities.h"
  13
  14
  15 //**********************************************************************************************************************
  16 vector<string> GetOtusCommand::setParameters(){
  17         try {
  18                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pgroup);
  19                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(plist);
  20                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
  21                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  22                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "GetOtusCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string GetOtusCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The get.otus command selects otus containing sequences from a specfic group or set of groups.\n";
  40                 helpString += "It outputs a new list file containing the otus containing sequences from in the those specified groups.\n";
  41                 helpString += "The get.otus command parameters are accnos, group, list, label and groups. The group and list parameters are required, unless you have valid current files.\n";
  42                 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";
  43                 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";
  44                 helpString += "The label parameter allows you to specify which distance you want to process.\n";
  45                 helpString += "The get.otus command should be in the following format: get.otus(accnos=yourAccnos, list=yourListFile, group=yourGroupFile, label=yourLabel).\n";
  46                 helpString += "Example get.otus(accnos=amazon.accnos, list=amazon.fn.list, group=amazon.groups, label=0.03).\n";
  47                 helpString += "or get.otus(groups=pasture, list=amazon.fn.list, amazon.groups, label=0.03).\n";
  48                 helpString += "Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListFile).\n";
  49                 return helpString;
  50         }
  51         catch(exception& e) {
  52                 m->errorOut(e, "GetOtusCommand", "getHelpString");
  53                 exit(1);
  54         }
  55 }
  56
  57 //**********************************************************************************************************************
  58 GetOtusCommand::GetOtusCommand(){
  59         try {
  60                 abort = true; calledHelp = true;
  61                 setParameters();
  62                 vector<string> tempOutNames;
  63                 outputTypes["group"] = tempOutNames;
  64                 outputTypes["list"] = tempOutNames;
  65         }
  66         catch(exception& e) {
  67                 m->errorOut(e, "GetOtusCommand", "GetOtusCommand");
  68                 exit(1);
  69         }
  70 }
  71 //**********************************************************************************************************************
  72 GetOtusCommand::GetOtusCommand(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, "GetOtusCommand", "GetOtusCommand");
 171                 exit(1);
 172         }
 173 }
 174 //**********************************************************************************************************************
 175
 176 int GetOtusCommand::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 get
 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> gNamesOfGroups = groupMap->getNamesOfGroups();
 191                 util->setGroups(Groups, gNamesOfGroups);
 192                 groupMap->setNamesOfGroups(gNamesOfGroups);
 193                 delete util;
 194
 195                 if (m->control_pressed) { delete groupMap; return 0; }
 196
 197                 //read through the list file keeping any otus that contain any sequence from the groups selected
 198                 readListGroup();
 199
 200                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 201
 202                 if (outputNames.size() != 0) {
 203                         m->mothurOutEndLine();
 204                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 205                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 206                         m->mothurOutEndLine();
 207
 208                         //set list file as new current listfile
 209                         string current = "";
 210                         itTypes = outputTypes.find("group");
 211                         if (itTypes != outputTypes.end()) {
 212                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 213                         }
 214
 215                         itTypes = outputTypes.find("list");
 216                         if (itTypes != outputTypes.end()) {
 217                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 218                         }
 219                 }
 220
 221                 return 0;
 222         }
 223
 224         catch(exception& e) {
 225                 m->errorOut(e, "GetOtusCommand", "execute");
 226                 exit(1);
 227         }
 228 }
 229 //**********************************************************************************************************************
 230 int GetOtusCommand::readListGroup(){
 231         try {
 232                 string thisOutputDir = outputDir;
 233                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 234                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick." + label +  m->getExtension(listfile);
 235
 236                 ofstream out;
 237                 m->openOutputFile(outputFileName, out);
 238
 239                 string GroupOutputDir = outputDir;
 240                 if (outputDir == "") {  GroupOutputDir += m->hasPath(groupfile);  }
 241                 string outputGroupFileName = GroupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick." + label  + m->getExtension(groupfile);
 242
 243                 ofstream outGroup;
 244                 m->openOutputFile(outputGroupFileName, outGroup);
 245
 246                 InputData* input = new InputData(listfile, "list");
 247                 ListVector* list = input->getListVector();
 248                 string lastLabel = list->getLabel();
 249
 250                 //using first label seen if none is provided
 251                 if (label == "") { label = lastLabel; }
 252
 253                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 254                 set<string> labels; labels.insert(label);
 255                 set<string> processedLabels;
 256                 set<string> userLabels = labels;
 257
 258                 bool wroteSomething = false;
 259
 260                 //as long as you are not at the end of the file or done wih the lines you want
 261                 while((list != NULL) && (userLabels.size() != 0)) {
 262
 263                         if (m->control_pressed) {  delete list; delete input; out.close();  outGroup.close(); m->mothurRemove(outputFileName);  m->mothurRemove(outputGroupFileName);return 0;  }
 264
 265                         if(labels.count(list->getLabel()) == 1){
 266                                 processList(list, groupMap, out, outGroup, wroteSomething);
 267
 268                                 processedLabels.insert(list->getLabel());
 269                                 userLabels.erase(list->getLabel());
 270                         }
 271
 272                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 273                                 string saveLabel = list->getLabel();
 274
 275                                 delete list;
 276
 277                                 list = input->getListVector(lastLabel);
 278
 279                                 processList(list, groupMap, out, outGroup, wroteSomething);
 280
 281                                 processedLabels.insert(list->getLabel());
 282                                 userLabels.erase(list->getLabel());
 283
 284                                 //restore real lastlabel to save below
 285                                 list->setLabel(saveLabel);
 286                         }
 287
 288                         lastLabel = list->getLabel();
 289
 290                         delete list; list = NULL;
 291
 292                         //get next line to process
 293                         list = input->getListVector();
 294                 }
 295
 296
 297                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); outGroup.close(); m->mothurRemove(outputFileName);  m->mothurRemove(outputGroupFileName); return 0;  }
 298
 299                 //output error messages about any remaining user labels
 300                 set<string>::iterator it;
 301                 bool needToRun = false;
 302                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 303                         m->mothurOut("Your file does not include the label " + *it);
 304                         if (processedLabels.count(lastLabel) != 1) {
 305                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 306                                 needToRun = true;
 307                         }else {
 308                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 309                         }
 310                 }
 311
 312                 //run last label if you need to
 313                 if (needToRun == true)  {
 314                         if (list != NULL) { delete list; }
 315
 316                         list = input->getListVector(lastLabel);
 317
 318                         processList(list, groupMap, out, outGroup, wroteSomething);
 319
 320                         delete list; list = NULL;
 321                 }
 322
 323                 out.close();
 324                 outGroup.close();
 325
 326                 if (wroteSomething == false) {  m->mothurOut("At distance " + label + " your file does NOT contain any otus containing sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 327                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 328                 outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName);
 329
 330                 return 0;
 331
 332         }
 333         catch(exception& e) {
 334                 m->errorOut(e, "GetOtusCommand", "readList");
 335                 exit(1);
 336         }
 337 }
 338 //**********************************************************************************************************************
 339 int GetOtusCommand::processList(ListVector*& list, GroupMap*& groupMap, ofstream& out, ofstream& outGroup, bool& wroteSomething){
 340         try {
 341
 342                 //make a new list vector
 343                 ListVector newList;
 344                 newList.setLabel(list->getLabel());
 345
 346                 int numOtus = 0;
 347                 //for each bin
 348                 for (int i = 0; i < list->getNumBins(); i++) {
 349                         if (m->control_pressed) { return 0; }
 350
 351                         //parse out names that are in accnos file
 352                         string binnames = list->get(i);
 353
 354                         bool keepBin = false;
 355                         string groupFileOutput = "";
 356
 357                         //parse names
 358                         string individual = "";
 359                         int length = binnames.length();
 360                         for(int j=0;j<length;j++){
 361                                 if(binnames[j] == ','){
 362                                         string group = groupMap->getGroup(individual);
 363                                         if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 364
 365                                         if (m->inUsersGroups(group, Groups)) {  keepBin = true; }
 366                                         groupFileOutput += individual + "\t" + group + "\n";
 367                                         individual = "";
 368
 369                                 }
 370                                 else{  individual += binnames[j];  }
 371                         }
 372
 373                         string group = groupMap->getGroup(individual);
 374                         if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 375
 376                         if (m->inUsersGroups(group, Groups)) {  keepBin = true; }
 377                         groupFileOutput += individual + "\t" + group + "\n";
 378
 379                         //if there are sequences from the groups we want in this bin add to new list, output to groupfile
 380                         if (keepBin) {
 381                                 newList.push_back(binnames);
 382                                 outGroup << groupFileOutput;
 383                                 numOtus++;
 384                         }
 385                 }
 386
 387                 //print new listvector
 388                 if (newList.getNumBins() != 0) {
 389                         wroteSomething = true;
 390                         newList.print(out);
 391                 }
 392
 393                 m->mothurOut(newList.getLabel() + " - selected " + toString(numOtus) + " of the " + toString(list->getNumBins()) + " OTUs."); m->mothurOutEndLine();
 394
 395                 return 0;
 396         }
 397         catch(exception& e) {
 398                 m->errorOut(e, "GetOtusCommand", "processList");
 399                 exit(1);
 400         }
 401 }
 402 //**********************************************************************************************************************
 403 void GetOtusCommand::readAccnos(){
 404         try {
 405                 Groups.clear();
 406
 407                 ifstream in;
 408                 m->openInputFile(accnosfile, in);
 409                 string name;
 410
 411                 while(!in.eof()){
 412                         in >> name;
 413
 414                         Groups.push_back(name);
 415
 416                         m->gobble(in);
 417                 }
 418                 in.close();
 419
 420         }
 421         catch(exception& e) {
 422                 m->errorOut(e, "GetOtusCommand", "readAccnos");
 423                 exit(1);
 424         }
 425 }
 426 //**********************************************************************************************************************
 427
 428