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
  17                 //allow user to run help
  18                 if(option == "help") {  help(); abort = true; }
  19
  20                 else {
  21                         //valid paramters for this command
  22                         string Array[] =  {"list","label","output","outputdir","inputdir"};
  23                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  24
  25                         OptionParser parser(option);
  26                         map<string,string> parameters = parser.getParameters();
  27
  28                         ValidParameters validParameter;
  29                         map<string,string>::iterator it;
  30
  31                         //check to make sure all parameters are valid for command
  32                         for (it = parameters.begin(); it != parameters.end(); it++) {
  33                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  34                         }
  35
  36                         //if the user changes the input directory command factory will send this info to us in the output parameter
  37                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  38                         if (inputDir == "not found"){   inputDir = "";          }
  39                         else {
  40                                 string path;
  41                                 it = parameters.find("list");
  42                                 //user has given a template file
  43                                 if(it != parameters.end()){
  44                                         path = hasPath(it->second);
  45                                         //if the user has not given a path then, add inputdir. else leave path alone.
  46                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
  47                                 }
  48                         }
  49
  50                         listFile = validParameter.validFile(parameters, "list", true);
  51                         if (listFile == "not found") { m->mothurOut("list is a required parameter for the otu.hierarchy command."); m->mothurOutEndLine(); abort = true; }
  52                         else if (listFile == "not open") { abort = true; }
  53
  54                         //if the user changes the output directory command factory will send this info to us in the output parameter
  55                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  56                                 outputDir = "";
  57                                 outputDir += hasPath(listFile); //if user entered a file with a path then preserve it
  58                         }
  59
  60                         //check for optional parameter and set defaults
  61                         // ...at some point should added some additional type checking...
  62                         label = validParameter.validFile(parameters, "label", false);
  63                         if (label == "not found") { m->mothurOut("label is a required parameter for the otu.hierarchy command."); m->mothurOutEndLine(); abort = true; }
  64                         else {
  65                                 splitAtDash(label, labels);
  66                                 if (labels.size() != 2) { m->mothurOut("You must provide 2 labels."); m->mothurOutEndLine(); abort = true; }
  67                         }
  68
  69                         output = validParameter.validFile(parameters, "output", false);                 if (output == "not found") { output = "name"; }
  70
  71                         if ((output != "name") && (output != "number")) { m->mothurOut("output options are name and number. I will use name."); m->mothurOutEndLine(); output = "name"; }
  72                 }
  73
  74         }
  75         catch(exception& e) {
  76                 m->errorOut(e, "OtuHierarchyCommand", "OtuHierarchyCommand");
  77                 exit(1);
  78         }
  79 }
  80 //**********************************************************************************************************************
  81
  82 void OtuHierarchyCommand::help(){
  83         try {
  84                 m->mothurOut("The otu.hierarchy command is used to see how otus relate at two distances. \n");
  85                 m->mothurOut("The otu.hierarchy command parameters are list, label and output.  list and label parameters are required. \n");
  86                 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");
  87                 m->mothurOut("The otu.hierarchy command should be in the following format: \n");
  88                 m->mothurOut("otu.hierarchy(list=yourListFile, label=yourLabels).\n");
  89                 m->mothurOut("Example otu.hierarchy(list=amazon.fn.list, label=0.01-0.03).\n");
  90                 m->mothurOut("The otu.hierarchy command outputs a .otu.hierarchy file which is described on the wiki.\n");
  91                 m->mothurOut("Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListFile).\n\n");
  92         }
  93         catch(exception& e) {
  94                 m->errorOut(e, "OtuHierarchyCommand", "help");
  95                 exit(1);
  96         }
  97 }
  98
  99 //**********************************************************************************************************************
 100
 101 OtuHierarchyCommand::~OtuHierarchyCommand(){}
 102
 103 //**********************************************************************************************************************
 104
 105 int OtuHierarchyCommand::execute(){
 106         try {
 107
 108                 if (abort == true) { return 0; }
 109
 110                 //get listvectors that correspond to labels requested, (or use smart distancing to get closest listvector)
 111                 vector<ListVector> lists = getListVectors();
 112
 113                 //determine which is little and which is big, putting little first
 114                 if (lists.size() == 2) {
 115                         //if big is first swap them
 116                         if (lists[0].getNumBins() < lists[1].getNumBins()) {
 117                                 reverse(lists.begin(), lists.end());
 118                         }
 119                 }else{
 120                         m->mothurOut("error getting listvectors, unable to read 2 different vectors, check your label inputs."); m->mothurOutEndLine(); return 0;
 121                 }
 122
 123                 //map sequences to bin number in the "little" otu
 124                 map<string, int> littleBins;
 125                 for (int i = 0; i < lists[0].getNumBins(); i++) {
 126                         string names = lists[0].get(i);
 127
 128                         //parse bin
 129                         while (names.find_first_of(',') != -1) {
 130                                 string name = names.substr(0,names.find_first_of(','));
 131                                 names = names.substr(names.find_first_of(',')+1, names.length());
 132                                 littleBins[name] = i;
 133                         }
 134
 135                         //get last name
 136                         littleBins[names] = i;
 137                 }
 138
 139                 ofstream out;
 140                 string outputFileName = outputDir + getRootName(getSimpleName(listFile)) + lists[0].getLabel() + "-" + lists[1].getLabel() + ".otu.hierarchy";
 141                 openOutputFile(outputFileName, out);
 142
 143                 //go through each bin in "big" otu and output the bins in "little" otu which created it
 144                 for (int i = 0; i < lists[1].getNumBins(); i++) {
 145
 146                         string names = lists[1].get(i);
 147
 148                         //output column 1
 149                         if (output == "name")   {   out << names << '\t';       }
 150                         else                                    {       out << i << '\t';               }
 151
 152                         map<int, int> bins; //bin numbers in little that are in this bin in big
 153                         map<int, int>::iterator it;
 154
 155                         //parse bin
 156                         while (names.find_first_of(',') != -1) {
 157                                 string name = names.substr(0,names.find_first_of(','));
 158                                 names = names.substr(names.find_first_of(',')+1, names.length());
 159                                 bins[littleBins[name]] = littleBins[name];
 160                         }
 161
 162                         //get last name
 163                         bins[littleBins[names]] = littleBins[names];
 164
 165                         string col2 = "";
 166                         for (it = bins.begin(); it != bins.end(); it++) {
 167                                 if (output == "name")   {   col2 += lists[0].get(it->first) + "\t";     }
 168                                 else                                    {       col2 += toString(it->first) + "\t";             }
 169                         }
 170
 171                         //output column 2
 172                         out << col2 << endl;
 173                 }
 174
 175                 out.close();
 176
 177                 m->mothurOutEndLine();
 178                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 179                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 180                 m->mothurOutEndLine();
 181
 182                 return 0;
 183         }
 184         catch(exception& e) {
 185                 m->errorOut(e, "OtuHierarchyCommand", "execute");
 186                 exit(1);
 187         }
 188 }
 189
 190 //**********************************************************************************************************************
 191 //returns a vector of listVectors where "little" vector is first
 192 vector<ListVector> OtuHierarchyCommand::getListVectors() {
 193         try {
 194
 195                 int pos; //to use in smart distancing, position of last read in file
 196                 int lastPos;
 197                 vector<ListVector> lists;
 198
 199                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 200                 set<string> processedLabels;
 201                 set<string> userLabels = labels;
 202
 203                 //open file
 204                 ifstream in;
 205                 openInputFile(listFile, in);
 206
 207                 //get first list vector in file
 208                 ListVector* list = NULL;
 209                 string lastLabel = "";
 210                 if (!in.eof())  {
 211                         pos = in.tellg();
 212                         lastPos = pos;
 213                         list = new ListVector(in);
 214                         gobble(in);
 215                         lastLabel = list->getLabel();
 216                 }
 217
 218                 while ((list != NULL) && (userLabels.size() != 0)) {
 219
 220                         //is this a listvector that we want?
 221                         if(labels.count(list->getLabel()) == 1){
 222
 223                                 //make copy of listvector
 224                                 ListVector temp(*list);
 225                                 lists.push_back(temp);
 226
 227                                 processedLabels.insert(list->getLabel());
 228                                 userLabels.erase(list->getLabel());
 229                         }
 230
 231                         //you have a label the user want that is smaller than this label and the last label has not already been processed
 232                         if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 233                                 string saveLabel = list->getLabel();
 234                                 int savePos = in.tellg();
 235
 236                                 //get smart distance line
 237                                 delete list;
 238                                 in.seekg(lastPos);
 239                                 if (!in.eof())  {
 240                                         list = new ListVector(in);
 241                                 }else { list = NULL; }
 242
 243                                 //make copy of listvector
 244                                 ListVector temp(*list);
 245                                 lists.push_back(temp);
 246
 247                                 processedLabels.insert(list->getLabel());
 248                                 userLabels.erase(list->getLabel());
 249
 250                                 //restore real lastlabel to save below
 251                                 list->setLabel(saveLabel);
 252                                 in.seekg(savePos);
 253                         }
 254
 255                         lastLabel = list->getLabel();
 256                         lastPos = pos;
 257
 258                         //get next line
 259                         delete list;
 260                         if (!in.eof())  {
 261                                 pos = in.tellg();
 262                                 list = new ListVector(in);
 263                                 gobble(in);
 264                         }else { list = NULL; }
 265                 }
 266
 267
 268
 269                 //output error messages about any remaining user labels
 270                 set<string>::iterator it;
 271                 bool needToRun = false;
 272                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 273                         m->mothurOut("Your file does not include the label " + *it);
 274                         if (processedLabels.count(lastLabel) != 1) {
 275                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 276                                 needToRun = true;
 277                         }else {
 278                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 279                         }
 280                 }
 281
 282                 //run last label if you need to
 283                 if (needToRun == true)  {
 284                         if (list != NULL) {     delete list;    }
 285
 286                         in.seekg(lastPos);
 287                         if (!in.eof())  {
 288                                 list = new ListVector(in);
 289
 290                                 //make copy of listvector
 291                                 ListVector temp(*list);
 292                                 lists.push_back(temp);
 293
 294                                 delete list;
 295                         }
 296                 }
 297
 298
 299                 return lists;
 300         }
 301         catch(exception& e) {
 302                 m->errorOut(e, "OtuHierarchyCommand", "getListVectors");
 303                 exit(1);
 304         }
 305 }
 306
 307 //**********************************************************************************************************************
 308
 309
 310
 311
 312