weighted.cpp

   1 /*
   2  *  weighted.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 2/9/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "weighted.h"
  11
  12 /**************************************************************************************************/
  13
  14 EstOutput Weighted::getValues(Tree* t, int p, string o) {
  15     try {
  16                 data.clear(); //clear out old values
  17                 int numGroups;
  18                 vector<double> D;
  19                 processors = p;
  20                 outputDir = o;
  21
  22         CountTable* ct = t->getCountTable();
  23
  24                 numGroups = m->getNumGroups();
  25
  26                 if (m->control_pressed) { return data; }
  27
  28                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
  29                 vector< vector<string> > namesOfGroupCombos;
  30                 for (int i=0; i<numGroups; i++) {
  31                         for (int l = 0; l < i; l++) {
  32                                 //initialize weighted scores
  33                                 //WScore[globaldata->Groups[i]+globaldata->Groups[l]] = 0.0;
  34                                 vector<string> groups; groups.push_back((m->getGroups())[i]); groups.push_back((m->getGroups())[l]);
  35                                 namesOfGroupCombos.push_back(groups);
  36                         }
  37                 }
  38
  39         int numPairs = namesOfGroupCombos.size();
  40         int numPairsPerProcessor = numPairs / processors;
  41
  42         for (int i = 0; i < processors; i++) {
  43             int startPos = i * numPairsPerProcessor;
  44             if(i == processors - 1){ numPairsPerProcessor = numPairs - i * numPairsPerProcessor; }
  45             lines.push_back(linePair(startPos, numPairsPerProcessor));
  46         }
  47
  48         data = createProcesses(t, namesOfGroupCombos, ct);
  49
  50         lines.clear();
  51
  52                 return data;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "Weighted", "getValues");
  56                 exit(1);
  57         }
  58 }
  59 /**************************************************************************************************/
  60
  61 EstOutput Weighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, CountTable* ct) {
  62         try {
  63         vector<int> processIDS;
  64                 EstOutput results;
  65 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
  66                 int process = 1;
  67
  68                 //loop through and create all the processes you want
  69                 while (process != processors) {
  70                         int pid = fork();
  71
  72                         if (pid > 0) {
  73                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
  74                                 process++;
  75                         }else if (pid == 0){
  76                                 EstOutput Myresults;
  77                                 Myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, ct);
  78
  79                                 //pass numSeqs to parent
  80                                 ofstream out;
  81                                 string tempFile = outputDir + toString(getpid()) + ".weighted.results.temp";
  82                                 m->openOutputFile(tempFile, out);
  83
  84                                 out << Myresults.size() << endl;
  85                                 for (int i = 0; i < Myresults.size(); i++) {  out << Myresults[i] << '\t';  } out << endl;
  86                                 out.close();
  87
  88                                 exit(0);
  89                         }else {
  90                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
  91                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
  92                                 exit(0);
  93                         }
  94                 }
  95
  96                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, ct);
  97
  98                 //force parent to wait until all the processes are done
  99                 for (int i=0;i<(processors-1);i++) {
 100                         int temp = processIDS[i];
 101                         wait(&temp);
 102                 }
 103
 104                 if (m->control_pressed) { return results; }
 105
 106                 //get data created by processes
 107                 for (int i=0;i<(processors-1);i++) {
 108                         ifstream in;
 109                         string s = outputDir + toString(processIDS[i]) + ".weighted.results.temp";
 110                         m->openInputFile(s, in);
 111
 112                         //get quantiles
 113                         while (!in.eof()) {
 114                                 int num;
 115                                 in >> num; m->gobble(in);
 116
 117                                 if (m->control_pressed) { break; }
 118
 119                                 double w;
 120                                 for (int j = 0; j < num; j++) {
 121                                         in >> w;
 122                                         results.push_back(w);
 123                                 }
 124                                 m->gobble(in);
 125                         }
 126                         in.close();
 127                         m->mothurRemove(s);
 128                 }
 129 #else
 130
 131         //fill in functions
 132         vector<weightedData*> pDataArray;
 133                 DWORD   dwThreadIdArray[processors-1];
 134                 HANDLE  hThreadArray[processors-1];
 135         vector<CountTable*> cts;
 136         vector<Tree*> trees;
 137
 138                 //Create processor worker threads.
 139                 for( int i=1; i<processors; i++ ){
 140             CountTable* copyCount = new CountTable();
 141             copyCount->copy(ct);
 142             Tree* copyTree = new Tree(copyCount);
 143             copyTree->getCopy(t);
 144
 145             cts.push_back(copyCount);
 146             trees.push_back(copyTree);
 147
 148             weightedData* tempweighted = new weightedData(m, lines[i].start, lines[i].num, namesOfGroupCombos, copyTree, copyCount, includeRoot);
 149                         pDataArray.push_back(tempweighted);
 150                         processIDS.push_back(i);
 151
 152                         hThreadArray[i-1] = CreateThread(NULL, 0, MyWeightedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 153                 }
 154
 155                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, ct);
 156
 157                 //Wait until all threads have terminated.
 158                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 159
 160                 //Close all thread handles and free memory allocations.
 161                 for(int i=0; i < pDataArray.size(); i++){
 162             for (int j = 0; j < pDataArray[i]->results.size(); j++) {  results.push_back(pDataArray[i]->results[j]);  }
 163                         delete cts[i];
 164             delete trees[i];
 165                         CloseHandle(hThreadArray[i]);
 166                         delete pDataArray[i];
 167                 }
 168 #endif
 169
 170         return results;
 171         }
 172         catch(exception& e) {
 173                 m->errorOut(e, "Weighted", "createProcesses");
 174                 exit(1);
 175         }
 176 }
 177 /**************************************************************************************************/
 178 EstOutput Weighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, CountTable* ct) {
 179  try {
 180                 EstOutput results;
 181                 vector<double> D;
 182
 183                 int count = 0;
 184                 for (int h = start; h < (start+num); h++) {
 185
 186                         if (m->control_pressed) { return results; }
 187
 188                         //initialize weighted score
 189                         string groupA = namesOfGroupCombos[h][0];
 190                         string groupB = namesOfGroupCombos[h][1];
 191
 192                         set<int> validBranches;
 193                         WScore[groupA+groupB] = 0.0;
 194                         D.push_back(0.0000); //initialize a spot in D for each combination
 195
 196                         //adding the wieghted sums from group i
 197                         for (int j = 0; j < t->groupNodeInfo[groupA].size(); j++) { //the leaf nodes that have seqs from group i
 198                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupA][j]].pcount.find(groupA);
 199                                 int numSeqsInGroupI = it->second;
 200
 201                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupA][j], groupA, groupB);
 202                                 double weightedSum = ((numSeqsInGroupI * sum) / (double)ct->getGroupCount(groupA));
 203
 204                                 D[count] += weightedSum;
 205                         }
 206
 207                         //adding the wieghted sums from group l
 208                         for (int j = 0; j < t->groupNodeInfo[groupB].size(); j++) { //the leaf nodes that have seqs from group l
 209                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupB][j]].pcount.find(groupB);
 210                                 int numSeqsInGroupL = it->second;
 211
 212                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupB][j], groupA, groupB);
 213                                 double weightedSum = ((numSeqsInGroupL * sum) / (double)ct->getGroupCount(groupB));
 214
 215                                 D[count] += weightedSum;
 216                         }
 217                         count++;
 218                 }
 219
 220                 //calculate u for the group comb
 221                 for (int h = start; h < (start+num); h++) {
 222                         //report progress
 223                         //m->mothurOut("Processing combo: " + toString(h)); m->mothurOutEndLine();
 224
 225                         string groupA = namesOfGroupCombos[h][0];
 226                         string groupB = namesOfGroupCombos[h][1];
 227
 228                         //calculate u for the group comb
 229                         for(int i=0;i<t->getNumNodes();i++){
 230
 231                                 if (m->control_pressed) { return data; }
 232
 233                                 double u;
 234                                 //int pcountSize = 0;
 235                                 //does this node have descendants from groupA
 236                                 it = t->tree[i].pcount.find(groupA);
 237                                 //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 238                                 if (it != t->tree[i].pcount.end()) {
 239                                         u = (double) t->tree[i].pcount[groupA] / (double) ct->getGroupCount(groupA);
 240                                 }else { u = 0.00; }
 241
 242
 243                                 //does this node have descendants from group l
 244                                 it = t->tree[i].pcount.find(groupB);
 245
 246                                 //if it does subtract their percentage from u
 247                                 if (it != t->tree[i].pcount.end()) {
 248                                         u -= (double) t->tree[i].pcount[groupB] / (double) ct->getGroupCount(groupB);
 249                                 }
 250
 251                                 if (includeRoot) {
 252                                         if (t->tree[i].getBranchLength() != -1) {
 253                                                 u = abs(u * t->tree[i].getBranchLength());
 254                                                 WScore[(groupA+groupB)] += u;
 255                                         }
 256                                 }else {
 257                                         //if this is not the root then add it
 258                                         if (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0) {
 259                                                 if (t->tree[i].getBranchLength() != -1) {
 260                                                         u = abs(u * t->tree[i].getBranchLength());
 261                                                         WScore[(groupA+groupB)] += u;
 262                                                 }
 263                                         }
 264                                 }
 265                         }
 266
 267                 }
 268
 269                 /********************************************************/
 270                 //calculate weighted score for the group combination
 271                 double UN;
 272                 count = 0;
 273                 for (int h = start; h < (start+num); h++) {
 274                         UN = (WScore[namesOfGroupCombos[h][0]+namesOfGroupCombos[h][1]] / D[count]);
 275                         if (isnan(UN) || isinf(UN)) { UN = 0; }
 276                         results.push_back(UN);
 277                         count++;
 278                 }
 279
 280                 return results;
 281         }
 282         catch(exception& e) {
 283                 m->errorOut(e, "Weighted", "driver");
 284                 exit(1);
 285         }
 286 }
 287 /**************************************************************************************************/
 288 EstOutput Weighted::getValues(Tree* t, string groupA, string groupB) {
 289  try {
 290
 291                 data.clear(); //clear out old values
 292
 293         CountTable* ct = t->getCountTable();
 294
 295                 if (m->control_pressed) { return data; }
 296
 297                 //initialize weighted score
 298                 WScore[(groupA+groupB)] = 0.0;
 299                 double D = 0.0;
 300                 set<int> validBranches;
 301
 302                 vector<string> groups; groups.push_back(groupA); groups.push_back(groupB);
 303
 304                 //adding the wieghted sums from group i
 305                 for (int j = 0; j < t->groupNodeInfo[groups[0]].size(); j++) { //the leaf nodes that have seqs from group i
 306                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[0]][j]].pcount.find(groups[0]);
 307                         int numSeqsInGroupI = it->second;
 308
 309                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[0]][j], groups[0], groups[1]);
 310                         double weightedSum = ((numSeqsInGroupI * sum) / (double)ct->getGroupCount(groups[0]));
 311
 312                         D += weightedSum;
 313                 }
 314
 315                 //adding the wieghted sums from group l
 316                 for (int j = 0; j < t->groupNodeInfo[groups[1]].size(); j++) { //the leaf nodes that have seqs from group l
 317                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[1]][j]].pcount.find(groups[1]);
 318                         int numSeqsInGroupL = it->second;
 319
 320                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[1]][j], groups[0], groups[1]);
 321                         double weightedSum = ((numSeqsInGroupL * sum) / (double)ct->getGroupCount(groups[1]));
 322
 323                         D += weightedSum;
 324                 }
 325
 326                 //calculate u for the group comb
 327                 for(int i=0;i<t->getNumNodes();i++){
 328
 329                         if (m->control_pressed) { return data; }
 330
 331                         double u;
 332                         //int pcountSize = 0;
 333                         //does this node have descendants from groupA
 334                         it = t->tree[i].pcount.find(groupA);
 335                         //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 336                         if (it != t->tree[i].pcount.end()) {
 337                                 u = (double) t->tree[i].pcount[groupA] / (double) ct->getGroupCount(groupA);
 338                         }else { u = 0.00; }
 339
 340
 341                         //does this node have descendants from group l
 342                         it = t->tree[i].pcount.find(groupB);
 343                         //if it does subtract their percentage from u
 344                         if (it != t->tree[i].pcount.end()) {
 345                                 u -= (double) t->tree[i].pcount[groupB] / (double) ct->getGroupCount(groupB);
 346                         }
 347
 348                         if (includeRoot) {
 349                                 if (t->tree[i].getBranchLength() != -1) {
 350                                         u = abs(u * t->tree[i].getBranchLength());
 351                                         WScore[(groupA+groupB)] += u;
 352                                 }
 353                         }else{
 354                                 //if this is not the root then add it
 355                                 if (rootForGrouping[groups].count(i) == 0) {
 356                                         if (t->tree[i].getBranchLength() != -1) {
 357                                                 u = abs(u * t->tree[i].getBranchLength());
 358                                                 WScore[(groupA+groupB)] += u;
 359                                         }
 360                                 }
 361                         }
 362                 }
 363                 /********************************************************/
 364
 365                 //calculate weighted score for the group combination
 366                 double UN;
 367                 UN = (WScore[(groupA+groupB)] / D);
 368
 369                 if (isnan(UN) || isinf(UN)) { UN = 0; }
 370                 data.push_back(UN);
 371
 372                 return data;
 373         }
 374         catch(exception& e) {
 375                 m->errorOut(e, "Weighted", "getValues");
 376                 exit(1);
 377         }
 378 }
 379 /**************************************************************************************************/
 380 double Weighted::getLengthToRoot(Tree* t, int v, string groupA, string groupB) {
 381         try {
 382
 383                 double sum = 0.0;
 384                 int index = v;
 385
 386                 //you are a leaf
 387                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 388                 double tempTotal = 0.0;
 389                 index = t->tree[index].getParent();
 390
 391                 vector<string> grouping; grouping.push_back(groupA); grouping.push_back(groupB);
 392
 393                 rootForGrouping[grouping].insert(index);
 394
 395                 //while you aren't at root
 396                 while(t->tree[index].getParent() != -1){
 397
 398                         if (m->control_pressed) {  return sum; }
 399
 400                         int parent = t->tree[index].getParent();
 401
 402                         if (includeRoot) { //add everyone
 403                                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 404                         }else {
 405
 406                                 //am I the root for this grouping? if so I want to stop "early"
 407                                 //does my sibling have descendants from the users groups?
 408                                 int lc = t->tree[parent].getLChild();
 409                                 int rc = t->tree[parent].getRChild();
 410
 411                                 int sib = lc;
 412                                 if (lc == index) { sib = rc; }
 413
 414                                 map<string, int>::iterator itGroup;
 415                                 int pcountSize = 0;
 416                                 itGroup = t->tree[sib].pcount.find(groupA);
 417                                 if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 418                                 itGroup = t->tree[sib].pcount.find(groupB);
 419                                 if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 420
 421                                 //if yes, I am not the root so add me
 422                                 if (pcountSize != 0) {
 423                                         if (t->tree[index].getBranchLength() != -1) {
 424                                                 sum += abs(t->tree[index].getBranchLength()) + tempTotal;
 425                                                 tempTotal = 0.0;
 426                                         }else {
 427                                                 sum += tempTotal;
 428                                                 tempTotal = 0.0;
 429                                         }
 430                                         rootForGrouping[grouping].clear();
 431                                         rootForGrouping[grouping].insert(parent);
 432                                 }else { //if no, I may be the root so add my br to tempTotal until I am proven innocent
 433                                         if (t->tree[index].getBranchLength() != -1) {
 434                                                 tempTotal += abs(t->tree[index].getBranchLength());
 435                                         }
 436                                 }
 437                         }
 438
 439                         index = parent;
 440                 }
 441
 442                 //get all nodes above the root to add so we don't add their u values above
 443                 index = *(rootForGrouping[grouping].begin());
 444
 445                 while(t->tree[index].getParent() != -1){
 446                         int parent = t->tree[index].getParent();
 447                         rootForGrouping[grouping].insert(parent);
 448                         index = parent;
 449                 }
 450                 return sum;
 451         }
 452         catch(exception& e) {
 453                 m->errorOut(e, "Weighted", "getBranchLengthSums");
 454                 exit(1);
 455         }
 456 }
 457 /**************************************************************************************************/
 458
 459