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