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