summarysharedcommand.h

   1 #ifndef SUMMARYSHAREDCOMMAND_H
   2 #define SUMMARYSHAREDCOMMAND_H
   3 /*
   4  *  summarysharedcommand.h
   5  *  Dotur
   6  *
   7  *  Created by Sarah Westcott on 1/2/09.
   8  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   9  *
  10  */
  11
  12
  13 #include "command.hpp"
  14 #include "sharedrabundvector.h"
  15 #include "inputdata.h"
  16 #include "calculator.h"
  17 #include "validcalculator.h"
  18 #include "sharedsobscollectsummary.h"
  19 #include "sharedchao1.h"
  20 #include "sharedace.h"
  21 #include "sharednseqs.h"
  22 #include "sharedjabund.h"
  23 #include "sharedsorabund.h"
  24 #include "sharedjclass.h"
  25 #include "sharedsorclass.h"
  26 #include "sharedjest.h"
  27 #include "sharedsorest.h"
  28 #include "sharedthetayc.h"
  29 #include "sharedthetan.h"
  30 #include "sharedkstest.h"
  31 #include "whittaker.h"
  32 #include "sharedochiai.h"
  33 #include "sharedanderbergs.h"
  34 #include "sharedkulczynski.h"
  35 #include "sharedkulczynskicody.h"
  36 #include "sharedlennon.h"
  37 #include "sharedmorisitahorn.h"
  38 #include "sharedbraycurtis.h"
  39 #include "sharedjackknife.h"
  40 #include "whittaker.h"
  41 #include "odum.h"
  42 #include "canberra.h"
  43 #include "structeuclidean.h"
  44 #include "structchord.h"
  45 #include "hellinger.h"
  46 #include "manhattan.h"
  47 #include "structpearson.h"
  48 #include "soergel.h"
  49 #include "spearman.h"
  50 #include "structkulczynski.h"
  51 #include "structchi2.h"
  52 #include "speciesprofile.h"
  53 #include "hamming.h"
  54 #include "gower.h"
  55 #include "memchi2.h"
  56 #include "memchord.h"
  57 #include "memeuclidean.h"
  58 #include "mempearson.h"
  59 #include "sharedjsd.h"
  60
  61 class SummarySharedCommand : public Command {
  62
  63 public:
  64         SummarySharedCommand(string);
  65         SummarySharedCommand();
  66         ~SummarySharedCommand() {}
  67
  68         vector<string> setParameters();
  69         string getCommandName()                 { return "summary.shared";                      }
  70         string getCommandCategory()             { return "OTU-Based Approaches";        }
  71
  72         string getHelpString();
  73     string getOutputPattern(string);
  74         string getCitation() { return "http://www.mothur.org/wiki/Summary.shared"; }
  75         string getDescription()         { return "generate a summary file containing calculator values for each line in the OTU data and for all possible comparisons between groups"; }
  76
  77         int execute();
  78         void help() { m->mothurOut(getHelpString()); }
  79
  80
  81 private:
  82         struct linePair {
  83                 int start;
  84                 int end;
  85         };
  86         vector<linePair> lines;
  87         vector<Calculator*> sumCalculators;
  88         InputData* input;
  89
  90         bool abort, allLines, mult, all, createPhylip, subsample;
  91         set<string> labels; //holds labels to be used
  92         string label, calc, groups, sharedfile, output;
  93         vector<string>  Estimators, Groups, outputNames;
  94         vector<SharedRAbundVector*> lookup;
  95         string format, outputDir;
  96         int numGroups, processors, subsampleSize, iters;
  97         int process(vector<SharedRAbundVector*>, string, string);
  98         int driver(vector<SharedRAbundVector*>, int, int, string, string, vector< vector<seqDist> >&);
  99     int printSims(ostream&, vector< vector<double> >&);
 100
 101 };
 102
 103 /**************************************************************************************************/
 104 //custom data structure for threads to use.
 105 //main process handling the calcs that can do more than 2 groups
 106 // This is passed by void pointer so it can be any data type
 107 // that can be passed using a single void pointer (LPVOID).
 108 struct summarySharedData {
 109     vector<SharedRAbundVector*> thisLookup;
 110     vector< vector<seqDist> > calcDists;
 111     vector<string>  Estimators;
 112         unsigned long long start;
 113         unsigned long long end;
 114         MothurOut* m;
 115         string sumFile;
 116     int count;
 117
 118         summarySharedData(){}
 119         summarySharedData(string sf, MothurOut* mout, unsigned long long st, unsigned long long en, vector<string> est, vector<SharedRAbundVector*> lu) {
 120                 sumFile = sf;
 121                 m = mout;
 122                 start = st;
 123                 end = en;
 124         Estimators = est;
 125         thisLookup = lu;
 126         count=0;
 127         }
 128 };
 129 /**************************************************************************************************/
 130 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 131 #else
 132 static DWORD WINAPI MySummarySharedThreadFunction(LPVOID lpParam){
 133         summarySharedData* pDataArray;
 134         pDataArray = (summarySharedData*)lpParam;
 135
 136         try {
 137
 138         vector<Calculator*> sumCalculators;
 139         ValidCalculators validCalculator;
 140         for (int i=0; i<pDataArray->Estimators.size(); i++) {
 141             if (validCalculator.isValidCalculator("sharedsummary", pDataArray->Estimators[i]) == true) {
 142                 if (pDataArray->Estimators[i] == "sharedsobs") {
 143                     sumCalculators.push_back(new SharedSobsCS());
 144                 }else if (pDataArray->Estimators[i] == "sharedchao") {
 145                     sumCalculators.push_back(new SharedChao1());
 146                 }else if (pDataArray->Estimators[i] == "sharedace") {
 147                     sumCalculators.push_back(new SharedAce());
 148                 }else if (pDataArray->Estimators[i] == "jabund") {
 149                     sumCalculators.push_back(new JAbund());
 150                 }else if (pDataArray->Estimators[i] == "sorabund") {
 151                     sumCalculators.push_back(new SorAbund());
 152                 }else if (pDataArray->Estimators[i] == "jclass") {
 153                     sumCalculators.push_back(new Jclass());
 154                 }else if (pDataArray->Estimators[i] == "sorclass") {
 155                     sumCalculators.push_back(new SorClass());
 156                 }else if (pDataArray->Estimators[i] == "jest") {
 157                     sumCalculators.push_back(new Jest());
 158                 }else if (pDataArray->Estimators[i] == "sorest") {
 159                     sumCalculators.push_back(new SorEst());
 160                 }else if (pDataArray->Estimators[i] == "thetayc") {
 161                     sumCalculators.push_back(new ThetaYC());
 162                 }else if (pDataArray->Estimators[i] == "thetan") {
 163                     sumCalculators.push_back(new ThetaN());
 164                 }else if (pDataArray->Estimators[i] == "kstest") {
 165                     sumCalculators.push_back(new KSTest());
 166                 }else if (pDataArray->Estimators[i] == "sharednseqs") {
 167                     sumCalculators.push_back(new SharedNSeqs());
 168                 }else if (pDataArray->Estimators[i] == "ochiai") {
 169                     sumCalculators.push_back(new Ochiai());
 170                 }else if (pDataArray->Estimators[i] == "anderberg") {
 171                     sumCalculators.push_back(new Anderberg());
 172                 }else if (pDataArray->Estimators[i] == "kulczynski") {
 173                     sumCalculators.push_back(new Kulczynski());
 174                 }else if (pDataArray->Estimators[i] == "kulczynskicody") {
 175                     sumCalculators.push_back(new KulczynskiCody());
 176                 }else if (pDataArray->Estimators[i] == "lennon") {
 177                     sumCalculators.push_back(new Lennon());
 178                 }else if (pDataArray->Estimators[i] == "morisitahorn") {
 179                     sumCalculators.push_back(new MorHorn());
 180                 }else if (pDataArray->Estimators[i] == "braycurtis") {
 181                     sumCalculators.push_back(new BrayCurtis());
 182                 }else if (pDataArray->Estimators[i] == "whittaker") {
 183                     sumCalculators.push_back(new Whittaker());
 184                 }else if (pDataArray->Estimators[i] == "odum") {
 185                     sumCalculators.push_back(new Odum());
 186                 }else if (pDataArray->Estimators[i] == "canberra") {
 187                     sumCalculators.push_back(new Canberra());
 188                 }else if (pDataArray->Estimators[i] == "structeuclidean") {
 189                     sumCalculators.push_back(new StructEuclidean());
 190                 }else if (pDataArray->Estimators[i] == "structchord") {
 191                     sumCalculators.push_back(new StructChord());
 192                 }else if (pDataArray->Estimators[i] == "hellinger") {
 193                     sumCalculators.push_back(new Hellinger());
 194                 }else if (pDataArray->Estimators[i] == "manhattan") {
 195                     sumCalculators.push_back(new Manhattan());
 196                 }else if (pDataArray->Estimators[i] == "structpearson") {
 197                     sumCalculators.push_back(new StructPearson());
 198                 }else if (pDataArray->Estimators[i] == "soergel") {
 199                     sumCalculators.push_back(new Soergel());
 200                 }else if (pDataArray->Estimators[i] == "spearman") {
 201                     sumCalculators.push_back(new Spearman());
 202                 }else if (pDataArray->Estimators[i] == "structkulczynski") {
 203                     sumCalculators.push_back(new StructKulczynski());
 204                 }else if (pDataArray->Estimators[i] == "speciesprofile") {
 205                     sumCalculators.push_back(new SpeciesProfile());
 206                 }else if (pDataArray->Estimators[i] == "hamming") {
 207                     sumCalculators.push_back(new Hamming());
 208                 }else if (pDataArray->Estimators[i] == "structchi2") {
 209                     sumCalculators.push_back(new StructChi2());
 210                 }else if (pDataArray->Estimators[i] == "gower") {
 211                     sumCalculators.push_back(new Gower());
 212                 }else if (pDataArray->Estimators[i] == "memchi2") {
 213                     sumCalculators.push_back(new MemChi2());
 214                 }else if (pDataArray->Estimators[i] == "memchord") {
 215                     sumCalculators.push_back(new MemChord());
 216                 }else if (pDataArray->Estimators[i] == "memeuclidean") {
 217                     sumCalculators.push_back(new MemEuclidean());
 218                 }else if (pDataArray->Estimators[i] == "mempearson") {
 219                     sumCalculators.push_back(new MemPearson());
 220                 }else if (pDataArray->Estimators[i] == "jsd") {
 221                     sumCalculators.push_back(new JSD());
 222                 }
 223             }
 224         }
 225
 226         pDataArray->calcDists.resize(sumCalculators.size());
 227
 228                 ofstream outputFileHandle;
 229                 pDataArray->m->openOutputFile(pDataArray->sumFile, outputFileHandle);
 230
 231                 vector<SharedRAbundVector*> subset;
 232                 for (int k = pDataArray->start; k < pDataArray->end; k++) { // pass cdd each set of groups to compare
 233             pDataArray->count++;
 234                         for (int l = 0; l < k; l++) {
 235
 236                                 outputFileHandle << pDataArray->thisLookup[0]->getLabel() << '\t';
 237
 238                                 subset.clear(); //clear out old pair of sharedrabunds
 239                                 //add new pair of sharedrabunds
 240                                 subset.push_back(pDataArray->thisLookup[k]); subset.push_back(pDataArray->thisLookup[l]);
 241
 242                                 //sort groups to be alphanumeric
 243                                 if (pDataArray->thisLookup[k]->getGroup() > pDataArray->thisLookup[l]->getGroup()) {
 244                                         outputFileHandle << (pDataArray->thisLookup[l]->getGroup() +'\t' + pDataArray->thisLookup[k]->getGroup()) << '\t'; //print out groups
 245                                 }else{
 246                                         outputFileHandle << (pDataArray->thisLookup[k]->getGroup() +'\t' + pDataArray->thisLookup[l]->getGroup()) << '\t'; //print out groups
 247                                 }
 248
 249                                 for(int i=0;i<sumCalculators.size();i++) {
 250
 251                                         //if this calc needs all groups to calculate the pair load all groups
 252                                         if (sumCalculators[i]->getNeedsAll()) {
 253                                                 //load subset with rest of lookup for those calcs that need everyone to calc for a pair
 254                                                 for (int w = 0; w < pDataArray->thisLookup.size(); w++) {
 255                                                         if ((w != k) && (w != l)) { subset.push_back(pDataArray->thisLookup[w]); }
 256                                                 }
 257                                         }
 258
 259                                         vector<double> tempdata = sumCalculators[i]->getValues(subset); //saves the calculator outputs
 260
 261                                         if (pDataArray->m->control_pressed) { for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; } outputFileHandle.close(); return 1; }
 262
 263                                         outputFileHandle << '\t';
 264                                         sumCalculators[i]->print(outputFileHandle);
 265
 266                                         seqDist temp(l, k, tempdata[0]);
 267                                         pDataArray->calcDists[i].push_back(temp);
 268                                 }
 269                                 outputFileHandle << endl;
 270                         }
 271                 }
 272
 273                 outputFileHandle.close();
 274         for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }
 275
 276                 return 0;
 277
 278         }
 279         catch(exception& e) {
 280                 pDataArray->m->errorOut(e, "SummarySharedCommand", "MySummarySharedThreadFunction");
 281                 exit(1);
 282         }
 283 }
 284 #endif
 285
 286
 287 #endif