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","output","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                         output = validParameter.validFile(parameters, "output", false);                 if (output == "not found") { output = "name"; }
  69
  70                         if ((output != "name") && (output != "number")) { mothurOut("output options are name and number. I will use name."); mothurOutEndLine(); output = "name"; }
  71                 }
  72
  73         }
  74         catch(exception& e) {
  75                 errorOut(e, "OtuHierarchyCommand", "OtuHierarchyCommand");
  76                 exit(1);
  77         }
  78 }
  79 //**********************************************************************************************************************
  80
  81 void OtuHierarchyCommand::help(){
  82         try {
  83                 mothurOut("The otu.hierarchy command is used to see how otus relate at two distances. \n");
  84                 mothurOut("The otu.hierarchy command parameters are list, label and output.  list and label parameters are required. \n");
  85                 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");
  86                 mothurOut("The otu.hierarchy command should be in the following format: \n");
  87                 mothurOut("otu.hierarchy(list=yourListFile, label=yourLabels).\n");
  88                 mothurOut("Example otu.hierarchy(list=amazon.fn.list, label=0.01-0.03).\n");
  89                 mothurOut("The otu.hierarchy command outputs a .otu.hierarchy file which is described on the wiki.\n");
  90                 mothurOut("Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListFile).\n\n");
  91         }
  92         catch(exception& e) {
  93                 errorOut(e, "OtuHierarchyCommand", "help");
  94                 exit(1);
  95         }
  96 }
  97
  98 //**********************************************************************************************************************
  99
 100 OtuHierarchyCommand::~OtuHierarchyCommand(){}
 101
 102 //**********************************************************************************************************************
 103
 104 int OtuHierarchyCommand::execute(){
 105         try {
 106
 107                 if (abort == true) { return 0; }
 108
 109                 //get listvectors that correspond to labels requested, (or use smart distancing to get closest listvector)
 110                 vector<ListVector> lists = getListVectors();
 111
 112                 //determine which is little and which is big, putting little first
 113                 if (lists.size() == 2) {
 114                         //if big is first swap them
 115                         if (lists[0].getNumBins() < lists[1].getNumBins()) {
 116                                 reverse(lists.begin(), lists.end());
 117                         }
 118                 }else{
 119                         mothurOut("error getting listvectors, unable to read 2 different vectors, check your label inputs."); mothurOutEndLine(); return 0;
 120                 }
 121
 122                 //map sequences to bin number in the "little" otu
 123                 map<string, int> littleBins;
 124                 for (int i = 0; i < lists[0].getNumBins(); i++) {
 125                         string names = lists[0].get(i);
 126
 127                         //parse bin
 128                         while (names.find_first_of(',') != -1) {
 129                                 string name = names.substr(0,names.find_first_of(','));
 130                                 names = names.substr(names.find_first_of(',')+1, names.length());
 131                                 littleBins[name] = i;
 132                         }
 133
 134                         //get last name
 135                         littleBins[names] = i;
 136                 }
 137
 138                 ofstream out;
 139                 string outputFileName = outputDir + getRootName(getSimpleName(listFile)) + lists[0].getLabel() + "-" + lists[1].getLabel() + ".otu.hierarchy";
 140                 openOutputFile(outputFileName, out);
 141
 142                 //go through each bin in "big" otu and output the bins in "little" otu which created it
 143                 for (int i = 0; i < lists[1].getNumBins(); i++) {
 144
 145                         string names = lists[1].get(i);
 146
 147                         //output column 1
 148                         if (output == "name")   {   out << names << '\t';       }
 149                         else                                    {       out << i << '\t';               }
 150
 151                         map<int, int> bins; //bin numbers in little that are in this bin in big
 152                         map<int, int>::iterator it;
 153
 154                         //parse bin
 155                         while (names.find_first_of(',') != -1) {
 156                                 string name = names.substr(0,names.find_first_of(','));
 157                                 names = names.substr(names.find_first_of(',')+1, names.length());
 158                                 bins[littleBins[name]] = littleBins[name];
 159                         }
 160
 161                         //get last name
 162                         bins[littleBins[names]] = littleBins[names];
 163
 164                         string col2 = "";
 165                         for (it = bins.begin(); it != bins.end(); it++) {
 166                                 if (output == "name")   {   col2 += lists[0].get(it->first) + "\t";     }
 167                                 else                                    {       col2 += toString(it->first) + "\t";             }
 168                         }
 169
 170                         //output column 2
 171                         out << col2 << endl;
 172                 }
 173
 174                 out.close();
 175
 176                 return 0;
 177         }
 178         catch(exception& e) {
 179                 errorOut(e, "OtuHierarchyCommand", "execute");
 180                 exit(1);
 181         }
 182 }
 183
 184 //**********************************************************************************************************************
 185 //returns a vector of listVectors where "little" vector is first
 186 vector<ListVector> OtuHierarchyCommand::getListVectors() {
 187         try {
 188
 189                 int pos; //to use in smart distancing, position of last read in file
 190                 int lastPos;
 191                 vector<ListVector> lists;
 192
 193                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 194                 set<string> processedLabels;
 195                 set<string> userLabels = labels;
 196
 197                 //open file
 198                 ifstream in;
 199                 openInputFile(listFile, in);
 200
 201                 //get first list vector in file
 202                 ListVector* list = NULL;
 203                 string lastLabel = "";
 204                 if (!in.eof())  {
 205                         pos = in.tellg();
 206                         lastPos = pos;
 207                         list = new ListVector(in);
 208                         gobble(in);
 209                         lastLabel = list->getLabel();
 210                 }
 211
 212                 while ((list != NULL) && (userLabels.size() != 0)) {
 213
 214                         //is this a listvector that we want?
 215                         if(labels.count(list->getLabel()) == 1){
 216
 217                                 //make copy of listvector
 218                                 ListVector temp(*list);
 219                                 lists.push_back(temp);
 220
 221                                 processedLabels.insert(list->getLabel());
 222                                 userLabels.erase(list->getLabel());
 223                         }
 224
 225                         //you have a label the user want that is smaller than this label and the last label has not already been processed
 226                         if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 227                                 string saveLabel = list->getLabel();
 228                                 int savePos = in.tellg();
 229
 230                                 //get smart distance line
 231                                 delete list;
 232                                 in.seekg(lastPos);
 233                                 if (!in.eof())  {
 234                                         list = new ListVector(in);
 235                                 }else { list = NULL; }
 236
 237                                 //make copy of listvector
 238                                 ListVector temp(*list);
 239                                 lists.push_back(temp);
 240
 241                                 processedLabels.insert(list->getLabel());
 242                                 userLabels.erase(list->getLabel());
 243
 244                                 //restore real lastlabel to save below
 245                                 list->setLabel(saveLabel);
 246                                 in.seekg(savePos);
 247                         }
 248
 249                         lastLabel = list->getLabel();
 250                         lastPos = pos;
 251
 252                         //get next line
 253                         delete list;
 254                         if (!in.eof())  {
 255                                 pos = in.tellg();
 256                                 list = new ListVector(in);
 257                                 gobble(in);
 258                         }else { list = NULL; }
 259                 }
 260
 261
 262
 263                 //output error messages about any remaining user labels
 264                 set<string>::iterator it;
 265                 bool needToRun = false;
 266                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 267                         mothurOut("Your file does not include the label " + *it);
 268                         if (processedLabels.count(lastLabel) != 1) {
 269                                 mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 270                                 needToRun = true;
 271                         }else {
 272                                 mothurOut(". Please refer to " + lastLabel + "."); mothurOutEndLine();
 273                         }
 274                 }
 275
 276                 //run last label if you need to
 277                 if (needToRun == true)  {
 278                         if (list != NULL) {     delete list;    }
 279
 280                         in.seekg(lastPos);
 281                         if (!in.eof())  {
 282                                 list = new ListVector(in);
 283
 284                                 //make copy of listvector
 285                                 ListVector temp(*list);
 286                                 lists.push_back(temp);
 287
 288                                 delete list;
 289                         }
 290                 }
 291
 292
 293                 return lists;
 294         }
 295         catch(exception& e) {
 296                 errorOut(e, "OtuHierarchyCommand", "getListVectors");
 297                 exit(1);
 298         }
 299 }
 300
 301 //**********************************************************************************************************************
 302
 303
 304
 305
 306