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","group",false,true, true); parameters.push_back(pgroup);
  19                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none","list",false,true, 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 string GetOtusCommand::getOutputPattern(string type) {
  58     try {
  59         string pattern = "";
  60
  61         if (type == "group")       {   pattern = "[filename],[tag],pick,[extension]";    }
  62         else if (type == "list")        {   pattern = "[filename],[tag],pick,[extension]";    }
  63         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  64
  65         return pattern;
  66     }
  67     catch(exception& e) {
  68         m->errorOut(e, "GetOtusCommand", "getOutputPattern");
  69         exit(1);
  70     }
  71 }
  72 //**********************************************************************************************************************
  73 GetOtusCommand::GetOtusCommand(){
  74         try {
  75                 abort = true; calledHelp = true;
  76                 setParameters();
  77                 vector<string> tempOutNames;
  78                 outputTypes["group"] = tempOutNames;
  79                 outputTypes["list"] = tempOutNames;
  80         }
  81         catch(exception& e) {
  82                 m->errorOut(e, "GetOtusCommand", "GetOtusCommand");
  83                 exit(1);
  84         }
  85 }
  86 //**********************************************************************************************************************
  87 GetOtusCommand::GetOtusCommand(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, "GetOtusCommand", "GetOtusCommand");
 186                 exit(1);
 187         }
 188 }
 189 //**********************************************************************************************************************
 190
 191 int GetOtusCommand::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 get
 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> gNamesOfGroups = groupMap->getNamesOfGroups();
 206                 util->setGroups(Groups, gNamesOfGroups);
 207                 groupMap->setNamesOfGroups(gNamesOfGroups);
 208                 delete util;
 209
 210                 if (m->control_pressed) { delete groupMap; return 0; }
 211
 212                 //read through the list file keeping any otus that contain any sequence from the groups selected
 213                 readListGroup();
 214
 215                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 216
 217                 if (outputNames.size() != 0) {
 218                         m->mothurOutEndLine();
 219                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 220                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 221                         m->mothurOutEndLine();
 222
 223                         //set list file as new current listfile
 224                         string current = "";
 225                         itTypes = outputTypes.find("group");
 226                         if (itTypes != outputTypes.end()) {
 227                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 228                         }
 229
 230                         itTypes = outputTypes.find("list");
 231                         if (itTypes != outputTypes.end()) {
 232                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 233                         }
 234                 }
 235
 236                 return 0;
 237         }
 238
 239         catch(exception& e) {
 240                 m->errorOut(e, "GetOtusCommand", "execute");
 241                 exit(1);
 242         }
 243 }
 244 //**********************************************************************************************************************
 245 int GetOtusCommand::readListGroup(){
 246         try {
 247                 string thisOutputDir = outputDir;
 248                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 249         map<string, string> variables;
 250         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
 251         variables["[tag]"] = label;
 252         variables["[extension]"] = m->getExtension(listfile);
 253                 string outputFileName = getOutputFileName("list", variables);
 254
 255                 ofstream out;
 256                 m->openOutputFile(outputFileName, out);
 257
 258                 string GroupOutputDir = outputDir;
 259                 if (outputDir == "") {  GroupOutputDir += m->hasPath(groupfile);  }
 260         variables["[filename]"] = GroupOutputDir + m->getRootName(m->getSimpleName(groupfile));
 261         variables["[extension]"] = m->getExtension(groupfile);
 262                 string outputGroupFileName = getOutputFileName("group", variables);
 263
 264                 ofstream outGroup;
 265                 m->openOutputFile(outputGroupFileName, outGroup);
 266
 267                 InputData* input = new InputData(listfile, "list");
 268                 ListVector* list = input->getListVector();
 269                 string lastLabel = list->getLabel();
 270
 271                 //using first label seen if none is provided
 272                 if (label == "") { label = lastLabel; }
 273
 274                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 275                 set<string> labels; labels.insert(label);
 276                 set<string> processedLabels;
 277                 set<string> userLabels = labels;
 278
 279                 bool wroteSomething = false;
 280
 281                 //as long as you are not at the end of the file or done wih the lines you want
 282                 while((list != NULL) && (userLabels.size() != 0)) {
 283
 284                         if (m->control_pressed) {  delete list; delete input; out.close();  outGroup.close(); m->mothurRemove(outputFileName);  m->mothurRemove(outputGroupFileName);return 0;  }
 285
 286                         if(labels.count(list->getLabel()) == 1){
 287                                 processList(list, groupMap, out, outGroup, wroteSomething);
 288
 289                                 processedLabels.insert(list->getLabel());
 290                                 userLabels.erase(list->getLabel());
 291                         }
 292
 293                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 294                                 string saveLabel = list->getLabel();
 295
 296                                 delete list;
 297
 298                                 list = input->getListVector(lastLabel);
 299
 300                                 processList(list, groupMap, out, outGroup, wroteSomething);
 301
 302                                 processedLabels.insert(list->getLabel());
 303                                 userLabels.erase(list->getLabel());
 304
 305                                 //restore real lastlabel to save below
 306                                 list->setLabel(saveLabel);
 307                         }
 308
 309                         lastLabel = list->getLabel();
 310
 311                         delete list; list = NULL;
 312
 313                         //get next line to process
 314                         list = input->getListVector();
 315                 }
 316
 317
 318                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); outGroup.close(); m->mothurRemove(outputFileName);  m->mothurRemove(outputGroupFileName); return 0;  }
 319
 320                 //output error messages about any remaining user labels
 321                 set<string>::iterator it;
 322                 bool needToRun = false;
 323                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 324                         m->mothurOut("Your file does not include the label " + *it);
 325                         if (processedLabels.count(lastLabel) != 1) {
 326                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 327                                 needToRun = true;
 328                         }else {
 329                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 330                         }
 331                 }
 332
 333                 //run last label if you need to
 334                 if (needToRun == true)  {
 335                         if (list != NULL) { delete list; }
 336
 337                         list = input->getListVector(lastLabel);
 338
 339                         processList(list, groupMap, out, outGroup, wroteSomething);
 340
 341                         delete list; list = NULL;
 342                 }
 343
 344                 out.close();
 345                 outGroup.close();
 346
 347                 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();  }
 348                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 349                 outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName);
 350
 351                 return 0;
 352
 353         }
 354         catch(exception& e) {
 355                 m->errorOut(e, "GetOtusCommand", "readList");
 356                 exit(1);
 357         }
 358 }
 359 //**********************************************************************************************************************
 360 int GetOtusCommand::processList(ListVector*& list, GroupMap*& groupMap, ofstream& out, ofstream& outGroup, bool& wroteSomething){
 361         try {
 362
 363                 //make a new list vector
 364                 ListVector newList;
 365                 newList.setLabel(list->getLabel());
 366
 367                 int numOtus = 0;
 368                 //for each bin
 369                 for (int i = 0; i < list->getNumBins(); i++) {
 370                         if (m->control_pressed) { return 0; }
 371
 372                         //parse out names that are in accnos file
 373                         string binnames = list->get(i);
 374
 375                         bool keepBin = false;
 376                         string groupFileOutput = "";
 377
 378                         //parse names
 379                         string individual = "";
 380                         int length = binnames.length();
 381                         for(int j=0;j<length;j++){
 382                                 if(binnames[j] == ','){
 383                                         string group = groupMap->getGroup(individual);
 384                                         if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 385
 386                                         if (m->inUsersGroups(group, Groups)) {  keepBin = true; }
 387                                         groupFileOutput += individual + "\t" + group + "\n";
 388                                         individual = "";
 389
 390                                 }
 391                                 else{  individual += binnames[j];  }
 392                         }
 393
 394                         string group = groupMap->getGroup(individual);
 395                         if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 396
 397                         if (m->inUsersGroups(group, Groups)) {  keepBin = true; }
 398                         groupFileOutput += individual + "\t" + group + "\n";
 399
 400                         //if there are sequences from the groups we want in this bin add to new list, output to groupfile
 401                         if (keepBin) {
 402                                 newList.push_back(binnames);
 403                                 outGroup << groupFileOutput;
 404                                 numOtus++;
 405                         }
 406                 }
 407
 408                 //print new listvector
 409                 if (newList.getNumBins() != 0) {
 410                         wroteSomething = true;
 411                         newList.print(out);
 412                 }
 413
 414                 m->mothurOut(newList.getLabel() + " - selected " + toString(numOtus) + " of the " + toString(list->getNumBins()) + " OTUs."); m->mothurOutEndLine();
 415
 416                 return 0;
 417         }
 418         catch(exception& e) {
 419                 m->errorOut(e, "GetOtusCommand", "processList");
 420                 exit(1);
 421         }
 422 }
 423 //**********************************************************************************************************************
 424
 425