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