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