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