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 #include "sharedrjsd.h"
  61
  62 class SummarySharedCommand : public Command {
  63
  64 public:
  65         SummarySharedCommand(string);
  66         SummarySharedCommand();
  67         ~SummarySharedCommand() {}
  68
  69         vector<string> setParameters();
  70         string getCommandName()                 { return "summary.shared";                      }
  71         string getCommandCategory()             { return "OTU-Based Approaches";        }
  72
  73         string getHelpString();
  74     string getOutputPattern(string);
  75         string getCitation() { return "http://www.mothur.org/wiki/Summary.shared"; }
  76         string getDescription()         { return "generate a summary file containing calculator values for each line in the OTU data and for all possible comparisons between groups"; }
  77
  78         int execute();
  79         void help() { m->mothurOut(getHelpString()); }
  80
  81
  82 private:
  83         struct linePair {
  84                 int start;
  85                 int end;
  86         };
  87         vector<linePair> lines;
  88         vector<Calculator*> sumCalculators;
  89         InputData* input;
  90
  91         bool abort, allLines, mult, all, createPhylip, subsample;
  92         set<string> labels; //holds labels to be used
  93         string label, calc, groups, sharedfile, output;
  94         vector<string>  Estimators, Groups, outputNames;
  95         vector<SharedRAbundVector*> lookup;
  96         string format, outputDir;
  97         int numGroups, processors, subsampleSize, iters;
  98         int process(vector<SharedRAbundVector*>, string, string);
  99         int driver(vector<SharedRAbundVector*>, int, int, string, string, vector< vector<seqDist> >&);
 100     int printSims(ostream&, vector< vector<double> >&);
 101
 102 };
 103
 104 /**************************************************************************************************/
 105 //custom data structure for threads to use.
 106 //main process handling the calcs that can do more than 2 groups
 107 // This is passed by void pointer so it can be any data type
 108 // that can be passed using a single void pointer (LPVOID).
 109 struct summarySharedData {
 110     vector<SharedRAbundVector*> thisLookup;
 111     vector< vector<seqDist> > calcDists;
 112     vector<string>  Estimators;
 113         unsigned long long start;
 114         unsigned long long end;
 115         MothurOut* m;
 116         string sumFile;
 117     int count;
 118
 119         summarySharedData(){}
 120         summarySharedData(string sf, MothurOut* mout, unsigned long long st, unsigned long long en, vector<string> est, vector<SharedRAbundVector*> lu) {
 121                 sumFile = sf;
 122                 m = mout;
 123                 start = st;
 124                 end = en;
 125         Estimators = est;
 126         thisLookup = lu;
 127         count=0;
 128         }
 129 };
 130 /**************************************************************************************************/
 131 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 132 #else
 133 static DWORD WINAPI MySummarySharedThreadFunction(LPVOID lpParam){
 134         summarySharedData* pDataArray;
 135         pDataArray = (summarySharedData*)lpParam;
 136
 137         try {
 138
 139         vector<Calculator*> sumCalculators;
 140         ValidCalculators validCalculator;
 141         for (int i=0; i<pDataArray->Estimators.size(); i++) {
 142             if (validCalculator.isValidCalculator("sharedsummary", pDataArray->Estimators[i]) == true) {
 143                 if (pDataArray->Estimators[i] == "sharedsobs") {
 144                     sumCalculators.push_back(new SharedSobsCS());
 145                 }else if (pDataArray->Estimators[i] == "sharedchao") {
 146                     sumCalculators.push_back(new SharedChao1());
 147                 }else if (pDataArray->Estimators[i] == "sharedace") {
 148                     sumCalculators.push_back(new SharedAce());
 149                 }else if (pDataArray->Estimators[i] == "jabund") {
 150                     sumCalculators.push_back(new JAbund());
 151                 }else if (pDataArray->Estimators[i] == "sorabund") {
 152                     sumCalculators.push_back(new SorAbund());
 153                 }else if (pDataArray->Estimators[i] == "jclass") {
 154                     sumCalculators.push_back(new Jclass());
 155                 }else if (pDataArray->Estimators[i] == "sorclass") {
 156                     sumCalculators.push_back(new SorClass());
 157                 }else if (pDataArray->Estimators[i] == "jest") {
 158                     sumCalculators.push_back(new Jest());
 159                 }else if (pDataArray->Estimators[i] == "sorest") {
 160                     sumCalculators.push_back(new SorEst());
 161                 }else if (pDataArray->Estimators[i] == "thetayc") {
 162                     sumCalculators.push_back(new ThetaYC());
 163                 }else if (pDataArray->Estimators[i] == "thetan") {
 164                     sumCalculators.push_back(new ThetaN());
 165                 }else if (pDataArray->Estimators[i] == "kstest") {
 166                     sumCalculators.push_back(new KSTest());
 167                 }else if (pDataArray->Estimators[i] == "sharednseqs") {
 168                     sumCalculators.push_back(new SharedNSeqs());
 169                 }else if (pDataArray->Estimators[i] == "ochiai") {
 170                     sumCalculators.push_back(new Ochiai());
 171                 }else if (pDataArray->Estimators[i] == "anderberg") {
 172                     sumCalculators.push_back(new Anderberg());
 173                 }else if (pDataArray->Estimators[i] == "kulczynski") {
 174                     sumCalculators.push_back(new Kulczynski());
 175                 }else if (pDataArray->Estimators[i] == "kulczynskicody") {
 176                     sumCalculators.push_back(new KulczynskiCody());
 177                 }else if (pDataArray->Estimators[i] == "lennon") {
 178                     sumCalculators.push_back(new Lennon());
 179                 }else if (pDataArray->Estimators[i] == "morisitahorn") {
 180                     sumCalculators.push_back(new MorHorn());
 181                 }else if (pDataArray->Estimators[i] == "braycurtis") {
 182                     sumCalculators.push_back(new BrayCurtis());
 183                 }else if (pDataArray->Estimators[i] == "whittaker") {
 184                     sumCalculators.push_back(new Whittaker());
 185                 }else if (pDataArray->Estimators[i] == "odum") {
 186                     sumCalculators.push_back(new Odum());
 187                 }else if (pDataArray->Estimators[i] == "canberra") {
 188                     sumCalculators.push_back(new Canberra());
 189                 }else if (pDataArray->Estimators[i] == "structeuclidean") {
 190                     sumCalculators.push_back(new StructEuclidean());
 191                 }else if (pDataArray->Estimators[i] == "structchord") {
 192                     sumCalculators.push_back(new StructChord());
 193                 }else if (pDataArray->Estimators[i] == "hellinger") {
 194                     sumCalculators.push_back(new Hellinger());
 195                 }else if (pDataArray->Estimators[i] == "manhattan") {
 196                     sumCalculators.push_back(new Manhattan());
 197                 }else if (pDataArray->Estimators[i] == "structpearson") {
 198                     sumCalculators.push_back(new StructPearson());
 199                 }else if (pDataArray->Estimators[i] == "soergel") {
 200                     sumCalculators.push_back(new Soergel());
 201                 }else if (pDataArray->Estimators[i] == "spearman") {
 202                     sumCalculators.push_back(new Spearman());
 203                 }else if (pDataArray->Estimators[i] == "structkulczynski") {
 204                     sumCalculators.push_back(new StructKulczynski());
 205                 }else if (pDataArray->Estimators[i] == "speciesprofile") {
 206                     sumCalculators.push_back(new SpeciesProfile());
 207                 }else if (pDataArray->Estimators[i] == "hamming") {
 208                     sumCalculators.push_back(new Hamming());
 209                 }else if (pDataArray->Estimators[i] == "structchi2") {
 210                     sumCalculators.push_back(new StructChi2());
 211                 }else if (pDataArray->Estimators[i] == "gower") {
 212                     sumCalculators.push_back(new Gower());
 213                 }else if (pDataArray->Estimators[i] == "memchi2") {
 214                     sumCalculators.push_back(new MemChi2());
 215                 }else if (pDataArray->Estimators[i] == "memchord") {
 216                     sumCalculators.push_back(new MemChord());
 217                 }else if (pDataArray->Estimators[i] == "memeuclidean") {
 218                     sumCalculators.push_back(new MemEuclidean());
 219                 }else if (pDataArray->Estimators[i] == "mempearson") {
 220                     sumCalculators.push_back(new MemPearson());
 221                 }else if (pDataArray->Estimators[i] == "jsd") {
 222                     sumCalculators.push_back(new JSD());
 223                 }else if (pDataArray->Estimators[i] == "rjsd") {
 224                     sumCalculators.push_back(new RJSD());
 225                 }
 226             }
 227         }
 228
 229         pDataArray->calcDists.resize(sumCalculators.size());
 230
 231                 ofstream outputFileHandle;
 232                 pDataArray->m->openOutputFile(pDataArray->sumFile, outputFileHandle);
 233
 234                 vector<SharedRAbundVector*> subset;
 235                 for (int k = pDataArray->start; k < pDataArray->end; k++) { // pass cdd each set of groups to compare
 236             pDataArray->count++;
 237                         for (int l = 0; l < k; l++) {
 238
 239                                 outputFileHandle << pDataArray->thisLookup[0]->getLabel() << '\t';
 240
 241                                 subset.clear(); //clear out old pair of sharedrabunds
 242                                 //add new pair of sharedrabunds
 243                                 subset.push_back(pDataArray->thisLookup[k]); subset.push_back(pDataArray->thisLookup[l]);
 244
 245                                 //sort groups to be alphanumeric
 246                                 if (pDataArray->thisLookup[k]->getGroup() > pDataArray->thisLookup[l]->getGroup()) {
 247                                         outputFileHandle << (pDataArray->thisLookup[l]->getGroup() +'\t' + pDataArray->thisLookup[k]->getGroup()) << '\t'; //print out groups
 248                                 }else{
 249                                         outputFileHandle << (pDataArray->thisLookup[k]->getGroup() +'\t' + pDataArray->thisLookup[l]->getGroup()) << '\t'; //print out groups
 250                                 }
 251
 252                                 for(int i=0;i<sumCalculators.size();i++) {
 253
 254                                         //if this calc needs all groups to calculate the pair load all groups
 255                                         if (sumCalculators[i]->getNeedsAll()) {
 256                                                 //load subset with rest of lookup for those calcs that need everyone to calc for a pair
 257                                                 for (int w = 0; w < pDataArray->thisLookup.size(); w++) {
 258                                                         if ((w != k) && (w != l)) { subset.push_back(pDataArray->thisLookup[w]); }
 259                                                 }
 260                                         }
 261
 262                                         vector<double> tempdata = sumCalculators[i]->getValues(subset); //saves the calculator outputs
 263
 264                                         if (pDataArray->m->control_pressed) { for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; } outputFileHandle.close(); return 1; }
 265
 266                                         outputFileHandle << '\t';
 267                                         sumCalculators[i]->print(outputFileHandle);
 268
 269                                         seqDist temp(l, k, tempdata[0]);
 270                                         pDataArray->calcDists[i].push_back(temp);
 271                                 }
 272                                 outputFileHandle << endl;
 273                         }
 274                 }
 275
 276                 outputFileHandle.close();
 277         for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }
 278
 279                 return 0;
 280
 281         }
 282         catch(exception& e) {
 283                 pDataArray->m->errorOut(e, "SummarySharedCommand", "MySummarySharedThreadFunction");
 284                 exit(1);
 285         }
 286 }
 287 #endif
 288
 289
 290 #endif