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 string GetOtusCommand::getOutputFileNameTag(string type, string inputName=""){
  58         try {
  59         string outputFileName = "";
  60                 map<string, vector<string> >::iterator it;
  61
  62         //is this a type this command creates
  63         it = outputTypes.find(type);
  64         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  65         else {
  66             if (type == "group")            {   outputFileName = "pick" + m->getExtension(inputName);  }
  67             else if (type == "list")        {   outputFileName = "pick" + m->getExtension(inputName);  }
  68             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  69         }
  70         return outputFileName;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "GetOtusCommand", "getOutputFileNameTag");
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78 GetOtusCommand::GetOtusCommand(){
  79         try {
  80                 abort = true; calledHelp = true;
  81                 setParameters();
  82                 vector<string> tempOutNames;
  83                 outputTypes["group"] = tempOutNames;
  84                 outputTypes["list"] = tempOutNames;
  85         }
  86         catch(exception& e) {
  87                 m->errorOut(e, "GetOtusCommand", "GetOtusCommand");
  88                 exit(1);
  89         }
  90 }
  91 //**********************************************************************************************************************
  92 GetOtusCommand::GetOtusCommand(string option)  {
  93         try {
  94                 abort = false; calledHelp = false;
  95
  96                 //allow user to run help
  97                 if(option == "help") { help(); abort = true; calledHelp = true; }
  98                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  99
 100                 else {
 101                         vector<string> myArray = setParameters();
 102
 103                         OptionParser parser(option);
 104                         map<string,string> parameters = parser.getParameters();
 105
 106                         ValidParameters validParameter;
 107                         map<string,string>::iterator it;
 108
 109                         //check to make sure all parameters are valid for command
 110                         for (it = parameters.begin(); it != parameters.end(); it++) {
 111                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 112                         }
 113
 114                         //initialize outputTypes
 115                         vector<string> tempOutNames;
 116                         outputTypes["group"] = tempOutNames;
 117                         outputTypes["list"] = tempOutNames;
 118
 119
 120                         //if the user changes the output directory command factory will send this info to us in the output parameter
 121                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 122
 123                         //if the user changes the input directory command factory will send this info to us in the output parameter
 124                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 125                         if (inputDir == "not found"){   inputDir = "";          }
 126                         else {
 127                                 string path;
 128                                 it = parameters.find("accnos");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 134                                 }
 135
 136                                 it = parameters.find("list");
 137                                 //user has given a template file
 138                                 if(it != parameters.end()){
 139                                         path = m->hasPath(it->second);
 140                                         //if the user has not given a path then, add inputdir. else leave path alone.
 141                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 142                                 }
 143
 144                                 it = parameters.find("group");
 145                                 //user has given a template file
 146                                 if(it != parameters.end()){
 147                                         path = m->hasPath(it->second);
 148                                         //if the user has not given a path then, add inputdir. else leave path alone.
 149                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 150                                 }
 151                         }
 152
 153
 154                         //check for required parameters
 155                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 156                         if (accnosfile == "not open") { abort = true; }
 157                         else if (accnosfile == "not found") {  accnosfile = ""; }
 158                         else { m->setAccnosFile(accnosfile); }
 159
 160                         groupfile = validParameter.validFile(parameters, "group", true);
 161                         if (groupfile == "not open") { abort = true; }
 162                         else if (groupfile == "not found") {
 163                                 groupfile = m->getGroupFile();
 164                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 165                                 else {  m->mothurOut("You have no current group file and the group parameter is required."); m->mothurOutEndLine(); abort = true; }
 166                         }else { m->setGroupFile(groupfile); }
 167
 168                         listfile = validParameter.validFile(parameters, "list", true);
 169                         if (listfile == "not open") { abort = true; }
 170                         else if (listfile == "not found") {
 171                                 listfile = m->getListFile();
 172                                 if (listfile != "") { m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 173                                 else {  m->mothurOut("You have no current list file and the list parameter is required."); m->mothurOutEndLine(); abort = true; }
 174                         }else { m->setListFile(listfile); }
 175
 176                         groups = validParameter.validFile(parameters, "groups", false);
 177                         if (groups == "not found") { groups = ""; }
 178                         else {
 179                                 m->splitAtDash(groups, Groups);
 180                         }
 181
 182                         label = validParameter.validFile(parameters, "label", false);
 183                         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=""; }
 184
 185                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 186                 }
 187
 188         }
 189         catch(exception& e) {
 190                 m->errorOut(e, "GetOtusCommand", "GetOtusCommand");
 191                 exit(1);
 192         }
 193 }
 194 //**********************************************************************************************************************
 195
 196 int GetOtusCommand::execute(){
 197         try {
 198
 199                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 200
 201                 groupMap = new GroupMap(groupfile);
 202                 groupMap->readMap();
 203
 204                 //get groups you want to get
 205                 if (accnosfile != "") { m->readAccnos(accnosfile, Groups); m->setGroups(Groups); }
 206
 207                 //make sure groups are valid
 208                 //takes care of user setting groupNames that are invalid or setting groups=all
 209                 SharedUtil* util = new SharedUtil();
 210                 vector<string> gNamesOfGroups = groupMap->getNamesOfGroups();
 211                 util->setGroups(Groups, gNamesOfGroups);
 212                 groupMap->setNamesOfGroups(gNamesOfGroups);
 213                 delete util;
 214
 215                 if (m->control_pressed) { delete groupMap; return 0; }
 216
 217                 //read through the list file keeping any otus that contain any sequence from the groups selected
 218                 readListGroup();
 219
 220                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 221
 222                 if (outputNames.size() != 0) {
 223                         m->mothurOutEndLine();
 224                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 225                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 226                         m->mothurOutEndLine();
 227
 228                         //set list file as new current listfile
 229                         string current = "";
 230                         itTypes = outputTypes.find("group");
 231                         if (itTypes != outputTypes.end()) {
 232                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 233                         }
 234
 235                         itTypes = outputTypes.find("list");
 236                         if (itTypes != outputTypes.end()) {
 237                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 238                         }
 239                 }
 240
 241                 return 0;
 242         }
 243
 244         catch(exception& e) {
 245                 m->errorOut(e, "GetOtusCommand", "execute");
 246                 exit(1);
 247         }
 248 }
 249 //**********************************************************************************************************************
 250 int GetOtusCommand::readListGroup(){
 251         try {
 252                 string thisOutputDir = outputDir;
 253                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 254                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + label + "." + getOutputFileNameTag("list", listfile);
 255
 256                 ofstream out;
 257                 m->openOutputFile(outputFileName, out);
 258
 259                 string GroupOutputDir = outputDir;
 260                 if (outputDir == "") {  GroupOutputDir += m->hasPath(groupfile);  }
 261                 string outputGroupFileName = GroupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + label + "." + getOutputFileNameTag("group", groupfile);
 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 does NOT contain any otus containing sequences from the groups you wish to get."); 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, "GetOtusCommand", "readList");
 355                 exit(1);
 356         }
 357 }
 358 //**********************************************************************************************************************
 359 int GetOtusCommand::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 keepBin = 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)) {  keepBin = true; }
 386                                         groupFileOutput += individual + "\t" + group + "\n";
 387                                         individual = "";
 388
 389                                 }
 390                                 else{  individual += binnames[j];  }
 391                         }
 392
 393                         string group = groupMap->getGroup(individual);
 394                         if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 395
 396                         if (m->inUsersGroups(group, Groups)) {  keepBin = true; }
 397                         groupFileOutput += individual + "\t" + group + "\n";
 398
 399                         //if there are sequences from the groups we want in this bin add to new list, output to groupfile
 400                         if (keepBin) {
 401                                 newList.push_back(binnames);
 402                                 outGroup << groupFileOutput;
 403                                 numOtus++;
 404                         }
 405                 }
 406
 407                 //print new listvector
 408                 if (newList.getNumBins() != 0) {
 409                         wroteSomething = true;
 410                         newList.print(out);
 411                 }
 412
 413                 m->mothurOut(newList.getLabel() + " - selected " + toString(numOtus) + " of the " + toString(list->getNumBins()) + " OTUs."); m->mothurOutEndLine();
 414
 415                 return 0;
 416         }
 417         catch(exception& e) {
 418                 m->errorOut(e, "GetOtusCommand", "processList");
 419                 exit(1);
 420         }
 421 }
 422 //**********************************************************************************************************************
 423
 424