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