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