otuhierarchycommand.cpp

   1 /*
   2  *  otuhierarchycommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 1/19/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "otuhierarchycommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> OtuHierarchyCommand::setParameters(){
  14         try {
  15                 CommandParameter poutput("output", "Multiple", "name-number", "name", "", "", "","",false,false); parameters.push_back(poutput);
  16                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none","otuheirarchy",false,true,true); parameters.push_back(plist);
  17                 CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
  18                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  19                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  20
  21                 vector<string> myArray;
  22                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  23                 return myArray;
  24         }
  25         catch(exception& e) {
  26                 m->errorOut(e, "OtuHierarchyCommand", "setParameters");
  27                 exit(1);
  28         }
  29 }
  30 //**********************************************************************************************************************
  31 string OtuHierarchyCommand::getHelpString(){
  32         try {
  33                 string helpString = "";
  34                 helpString += "The otu.hierarchy command is used to see how otus relate at two distances. \n";
  35                 helpString += "The otu.hierarchy command parameters are list, label and output.  list and label parameters are required. \n";
  36                 helpString += "The output parameter allows you to output the names of the sequence in the OTUs or the OTU numbers. Options are name and number, default is name. \n";
  37                 helpString += "The otu.hierarchy command should be in the following format: \n";
  38                 helpString += "otu.hierarchy(list=yourListFile, label=yourLabels).\n";
  39                 helpString += "Example otu.hierarchy(list=amazon.fn.list, label=0.01-0.03).\n";
  40                 helpString += "The otu.hierarchy command outputs a .otu.hierarchy file which is described on the wiki.\n";
  41                 helpString += "Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListFile).\n";
  42                 return helpString;
  43         }
  44         catch(exception& e) {
  45                 m->errorOut(e, "OtuHierarchyCommand", "getHelpString");
  46                 exit(1);
  47         }
  48 }
  49 //**********************************************************************************************************************
  50 string OtuHierarchyCommand::getOutputPattern(string type) {
  51     try {
  52         string pattern = "";
  53
  54         if (type == "otuheirarchy") {  pattern = "[filename],[distance1],[tag],[distance2],otu.hierarchy"; }
  55         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  56
  57         return pattern;
  58     }
  59     catch(exception& e) {
  60         m->errorOut(e, "OtuHierarchyCommand", "getOutputPattern");
  61         exit(1);
  62     }
  63 }
  64 //**********************************************************************************************************************
  65 OtuHierarchyCommand::OtuHierarchyCommand(){
  66         try {
  67                 abort = true; calledHelp = true;
  68                 setParameters();
  69                 vector<string> tempOutNames;
  70                 outputTypes["otuheirarchy"] = tempOutNames;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "OtuHierarchyCommand", "OtuHierarchyCommand");
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78 OtuHierarchyCommand::OtuHierarchyCommand(string option) {
  79         try {
  80                 abort = false; calledHelp = false;
  81
  82                 //allow user to run help
  83                 if(option == "help") {  help(); abort = true; calledHelp = true; }
  84                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  85
  86                 else {
  87                         vector<string> myArray = setParameters();
  88
  89                         OptionParser parser(option);
  90                         map<string,string> parameters = parser.getParameters();
  91
  92                         ValidParameters validParameter;
  93                         map<string,string>::iterator it;
  94
  95                         //check to make sure all parameters are valid for command
  96                         for (it = parameters.begin(); it != parameters.end(); it++) {
  97                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  98                         }
  99
 100                         //initialize outputTypes
 101                         vector<string> tempOutNames;
 102                         outputTypes["otuheirarchy"] = tempOutNames;
 103
 104                         //if the user changes the input directory command factory will send this info to us in the output parameter
 105                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 106                         if (inputDir == "not found"){   inputDir = "";          }
 107                         else {
 108                                 string path;
 109                                 it = parameters.find("list");
 110                                 //user has given a template file
 111                                 if(it != parameters.end()){
 112                                         path = m->hasPath(it->second);
 113                                         //if the user has not given a path then, add inputdir. else leave path alone.
 114                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 115                                 }
 116                         }
 117
 118                         listFile = validParameter.validFile(parameters, "list", true);
 119                         if (listFile == "not found") {
 120                                 listFile = m->getListFile();
 121                                 if (listFile != "") {  m->mothurOut("Using " + listFile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 122                                 else {
 123                                         m->mothurOut("No valid current list file. You must provide a list file."); m->mothurOutEndLine();
 124                                         abort = true;
 125                                 }
 126                         }else if (listFile == "not open") { abort = true; }
 127                         else { m->setListFile(listFile); }
 128
 129                         //if the user changes the output directory command factory will send this info to us in the output parameter
 130                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 131                                 outputDir = "";
 132                                 outputDir += m->hasPath(listFile); //if user entered a file with a path then preserve it
 133                         }
 134
 135                         //check for optional parameter and set defaults
 136                         // ...at some point should added some additional type checking...
 137                         label = validParameter.validFile(parameters, "label", false);
 138                         if (label == "not found") { m->mothurOut("label is a required parameter for the otu.hierarchy command."); m->mothurOutEndLine(); abort = true; }
 139                         else {
 140                                 m->splitAtDash(label, labels);
 141                                 if (labels.size() != 2) { m->mothurOut("You must provide 2 labels."); m->mothurOutEndLine(); abort = true; }
 142                         }
 143
 144                         output = validParameter.validFile(parameters, "output", false);                 if (output == "not found") { output = "name"; }
 145
 146                         if ((output != "name") && (output != "number")) { m->mothurOut("output options are name and number. I will use name."); m->mothurOutEndLine(); output = "name"; }
 147                 }
 148
 149         }
 150         catch(exception& e) {
 151                 m->errorOut(e, "OtuHierarchyCommand", "OtuHierarchyCommand");
 152                 exit(1);
 153         }
 154 }
 155 //**********************************************************************************************************************
 156
 157 int OtuHierarchyCommand::execute(){
 158         try {
 159
 160                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 161
 162                 //get listvectors that correspond to labels requested, (or use smart distancing to get closest listvector)
 163                 vector<ListVector> lists = getListVectors();
 164
 165                 if (m->control_pressed) { outputTypes.clear(); return 0; }
 166
 167                 //determine which is little and which is big, putting little first
 168                 if (lists.size() == 2) {
 169                         //if big is first swap them
 170                         if (lists[0].getNumBins() < lists[1].getNumBins()) {
 171                                 reverse(lists.begin(), lists.end());
 172                         }
 173                 }else{
 174                         m->mothurOut("error getting listvectors, unable to read 2 different vectors, check your label inputs."); m->mothurOutEndLine(); return 0;
 175                 }
 176
 177                 //map sequences to bin number in the "little" otu
 178                 map<string, int> littleBins;
 179                 for (int i = 0; i < lists[0].getNumBins(); i++) {
 180
 181                         if (m->control_pressed) {  return 0; }
 182
 183                         string names = lists[0].get(i);
 184
 185                         //parse bin
 186                         while (names.find_first_of(',') != -1) {
 187                                 string name = names.substr(0,names.find_first_of(','));
 188                                 names = names.substr(names.find_first_of(',')+1, names.length());
 189                                 littleBins[name] = i;
 190                         }
 191
 192                         //get last name
 193                         littleBins[names] = i;
 194                 }
 195
 196                 ofstream out;
 197         map<string, string> variables;
 198         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(listFile));
 199         variables["[distance1]"] = lists[0].getLabel();
 200         variables["[tag]"] = "-";
 201         variables["[distance2]"] = lists[1].getLabel();
 202                 string outputFileName = getOutputFileName("otuheirarchy",variables);
 203                 m->openOutputFile(outputFileName, out);
 204
 205                 //go through each bin in "big" otu and output the bins in "little" otu which created it
 206                 for (int i = 0; i < lists[1].getNumBins(); i++) {
 207
 208                         if (m->control_pressed) { outputTypes.clear(); out.close(); m->mothurRemove(outputFileName); return 0; }
 209
 210                         string names = lists[1].get(i);
 211
 212                         //output column 1
 213                         if (output == "name")   {   out << names << '\t';       }
 214                         else                                    {       out << i << '\t';               }
 215
 216                         map<int, int> bins; //bin numbers in little that are in this bin in big
 217                         map<int, int>::iterator it;
 218
 219                         //parse bin
 220                         while (names.find_first_of(',') != -1) {
 221                                 string name = names.substr(0,names.find_first_of(','));
 222                                 names = names.substr(names.find_first_of(',')+1, names.length());
 223                                 bins[littleBins[name]] = littleBins[name];
 224                         }
 225
 226                         //get last name
 227                         bins[littleBins[names]] = littleBins[names];
 228
 229                         string col2 = "";
 230                         for (it = bins.begin(); it != bins.end(); it++) {
 231                                 if (output == "name")   {   col2 += lists[0].get(it->first) + "\t";     }
 232                                 else                                    {       col2 += toString(it->first) + "\t";             }
 233                         }
 234
 235                         //output column 2
 236                         out << col2 << endl;
 237                 }
 238
 239                 out.close();
 240
 241                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outputFileName); return 0; }
 242
 243                 m->mothurOutEndLine();
 244                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 245                 m->mothurOut(outputFileName); m->mothurOutEndLine();    outputNames.push_back(outputFileName); outputTypes["otuheirarchy"].push_back(outputFileName);
 246                 m->mothurOutEndLine();
 247
 248                 return 0;
 249         }
 250         catch(exception& e) {
 251                 m->errorOut(e, "OtuHierarchyCommand", "execute");
 252                 exit(1);
 253         }
 254 }
 255
 256 //**********************************************************************************************************************
 257 //returns a vector of listVectors where "little" vector is first
 258 vector<ListVector> OtuHierarchyCommand::getListVectors() {
 259         try {
 260
 261                 int pos; //to use in smart distancing, position of last read in file
 262                 int lastPos;
 263                 vector<ListVector> lists;
 264
 265                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 266                 set<string> processedLabels;
 267                 set<string> userLabels = labels;
 268
 269                 //open file
 270                 ifstream in;
 271                 m->openInputFile(listFile, in);
 272
 273                 //get first list vector in file
 274                 ListVector* list = NULL;
 275                 string lastLabel = "";
 276                 if (!in.eof())  {
 277                         pos = in.tellg();
 278                         lastPos = pos;
 279                         list = new ListVector(in);
 280                         m->gobble(in);
 281                         lastLabel = list->getLabel();
 282                 }
 283
 284                 while ((list != NULL) && (userLabels.size() != 0)) {
 285
 286                         if (m->control_pressed) {  in.close(); delete list;  return lists; }
 287
 288                         //is this a listvector that we want?
 289                         if(labels.count(list->getLabel()) == 1){
 290
 291                                 //make copy of listvector
 292                                 ListVector temp(*list);
 293                                 lists.push_back(temp);
 294
 295                                 processedLabels.insert(list->getLabel());
 296                                 userLabels.erase(list->getLabel());
 297                         }
 298
 299                         //you have a label the user want that is smaller than this label and the last label has not already been processed
 300                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 301                                 string saveLabel = list->getLabel();
 302                                 int savePos = in.tellg();
 303
 304                                 //get smart distance line
 305                                 delete list;
 306                                 in.seekg(lastPos);
 307                                 if (!in.eof())  {
 308                                         list = new ListVector(in);
 309                                 }else { list = NULL; }
 310
 311                                 //make copy of listvector
 312                                 ListVector temp(*list);
 313                                 lists.push_back(temp);
 314
 315                                 processedLabels.insert(list->getLabel());
 316                                 userLabels.erase(list->getLabel());
 317
 318                                 //restore real lastlabel to save below
 319                                 list->setLabel(saveLabel);
 320                                 in.seekg(savePos);
 321                         }
 322
 323                         lastLabel = list->getLabel();
 324                         lastPos = pos;
 325
 326                         //get next line
 327                         delete list;
 328                         if (!in.eof())  {
 329                                 pos = in.tellg();
 330                                 list = new ListVector(in);
 331                                 m->gobble(in);
 332                         }else { list = NULL; }
 333                 }
 334
 335                 if (m->control_pressed) { in.close();  return lists; }
 336
 337                 //output error messages about any remaining user labels
 338                 set<string>::iterator it;
 339                 bool needToRun = false;
 340                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 341                         m->mothurOut("Your file does not include the label " + *it);
 342                         if (processedLabels.count(lastLabel) != 1) {
 343                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 344                                 needToRun = true;
 345                         }else {
 346                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 347                         }
 348                 }
 349
 350                 if (m->control_pressed) {  in.close(); return lists; }
 351
 352                 //run last label if you need to
 353                 if (needToRun == true)  {
 354                         if (list != NULL) {     delete list;    }
 355
 356                         in.seekg(lastPos);
 357                         if (!in.eof())  {
 358                                 list = new ListVector(in);
 359
 360                                 //make copy of listvector
 361                                 ListVector temp(*list);
 362                                 lists.push_back(temp);
 363
 364                                 delete list;
 365                         }
 366                 }
 367
 368                 in.close();
 369                 return lists;
 370         }
 371         catch(exception& e) {
 372                 m->errorOut(e, "OtuHierarchyCommand", "getListVectors");
 373                 exit(1);
 374         }
 375 }
 376
 377 //**********************************************************************************************************************
 378
 379
 380
 381
 382