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 remainingPairs = namesOfGroupCombos.size();
  40         int startIndex = 0;
  41         for (int remainingProcessors = processors; remainingProcessors > 0; remainingProcessors--) {
  42             int numPairs = remainingPairs; //case for last processor
  43             if (remainingProcessors != 1) { numPairs = ceil(remainingPairs / remainingProcessors); }
  44             lines.push_back(linePair(startIndex, numPairs)); //startIndex, numPairs
  45             startIndex = startIndex + numPairs;
  46             remainingPairs = remainingPairs - numPairs;
  47         }
  48
  49         data = createProcesses(t, namesOfGroupCombos, ct);
  50
  51         lines.clear();
  52
  53                 return data;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "Weighted", "getValues");
  57                 exit(1);
  58         }
  59 }
  60 /**************************************************************************************************/
  61
  62 EstOutput Weighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, CountTable* ct) {
  63         try {
  64         vector<int> processIDS;
  65                 EstOutput results;
  66 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
  67                 int process = 1;
  68
  69                 //loop through and create all the processes you want
  70                 while (process != processors) {
  71                         int pid = fork();
  72
  73                         if (pid > 0) {
  74                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
  75                                 process++;
  76                         }else if (pid == 0){
  77                                 EstOutput Myresults;
  78                                 Myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, ct);
  79
  80                                 //pass numSeqs to parent
  81                                 ofstream out;
  82                                 string tempFile = outputDir + toString(getpid()) + ".weighted.results.temp";
  83                                 m->openOutputFile(tempFile, out);
  84
  85                                 out << Myresults.size() << endl;
  86                                 for (int i = 0; i < Myresults.size(); i++) {  out << Myresults[i] << '\t';  } out << endl;
  87                                 out.close();
  88
  89                                 exit(0);
  90                         }else {
  91                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
  92                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
  93                                 exit(0);
  94                         }
  95                 }
  96
  97                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, ct);
  98
  99                 //force parent to wait until all the processes are done
 100                 for (int i=0;i<(processors-1);i++) {
 101                         int temp = processIDS[i];
 102                         wait(&temp);
 103                 }
 104
 105                 if (m->control_pressed) { return results; }
 106
 107                 //get data created by processes
 108                 for (int i=0;i<(processors-1);i++) {
 109                         ifstream in;
 110                         string s = outputDir + toString(processIDS[i]) + ".weighted.results.temp";
 111                         m->openInputFile(s, in);
 112
 113                         //get quantiles
 114                         while (!in.eof()) {
 115                                 int num;
 116                                 in >> num; m->gobble(in);
 117
 118                                 if (m->control_pressed) { break; }
 119
 120                                 double w;
 121                                 for (int j = 0; j < num; j++) {
 122                                         in >> w;
 123                                         results.push_back(w);
 124                                 }
 125                                 m->gobble(in);
 126                         }
 127                         in.close();
 128                         m->mothurRemove(s);
 129                 }
 130 #else
 131
 132         //fill in functions
 133         vector<weightedData*> pDataArray;
 134                 DWORD   dwThreadIdArray[processors-1];
 135                 HANDLE  hThreadArray[processors-1];
 136         vector<CountTable*> cts;
 137         vector<Tree*> trees;
 138
 139                 //Create processor worker threads.
 140                 for( int i=1; i<processors; i++ ){
 141             CountTable* copyCount = new CountTable();
 142             copyCount->copy(ct);
 143             Tree* copyTree = new Tree(copyCount);
 144             copyTree->getCopy(t);
 145
 146             cts.push_back(copyCount);
 147             trees.push_back(copyTree);
 148
 149             weightedData* tempweighted = new weightedData(m, lines[i].start, lines[i].num, namesOfGroupCombos, copyTree, copyCount, includeRoot);
 150                         pDataArray.push_back(tempweighted);
 151                         processIDS.push_back(i);
 152
 153                         hThreadArray[i-1] = CreateThread(NULL, 0, MyWeightedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 154                 }
 155
 156                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, ct);
 157
 158                 //Wait until all threads have terminated.
 159                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 160
 161                 //Close all thread handles and free memory allocations.
 162                 for(int i=0; i < pDataArray.size(); i++){
 163             for (int j = 0; j < pDataArray[i]->results.size(); j++) {  results.push_back(pDataArray[i]->results[j]);  }
 164                         delete cts[i];
 165             delete trees[i];
 166                         CloseHandle(hThreadArray[i]);
 167                         delete pDataArray[i];
 168                 }
 169 #endif
 170
 171         return results;
 172         }
 173         catch(exception& e) {
 174                 m->errorOut(e, "Weighted", "createProcesses");
 175                 exit(1);
 176         }
 177 }
 178 /**************************************************************************************************/
 179 EstOutput Weighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, CountTable* ct) {
 180  try {
 181                 EstOutput results;
 182                 vector<double> D;
 183
 184                 int count = 0;
 185                 for (int h = start; h < (start+num); h++) {
 186
 187                         if (m->control_pressed) { return results; }
 188
 189                         //initialize weighted score
 190                         string groupA = namesOfGroupCombos[h][0];
 191                         string groupB = namesOfGroupCombos[h][1];
 192
 193                         set<int> validBranches;
 194                         WScore[groupA+groupB] = 0.0;
 195                         D.push_back(0.0000); //initialize a spot in D for each combination
 196
 197                         //adding the wieghted sums from group i
 198                         for (int j = 0; j < t->groupNodeInfo[groupA].size(); j++) { //the leaf nodes that have seqs from group i
 199                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupA][j]].pcount.find(groupA);
 200                                 int numSeqsInGroupI = it->second;
 201
 202                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupA][j], groupA, groupB);
 203                                 double weightedSum = ((numSeqsInGroupI * sum) / (double)ct->getGroupCount(groupA));
 204
 205                                 D[count] += weightedSum;
 206                         }
 207
 208                         //adding the wieghted sums from group l
 209                         for (int j = 0; j < t->groupNodeInfo[groupB].size(); j++) { //the leaf nodes that have seqs from group l
 210                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupB][j]].pcount.find(groupB);
 211                                 int numSeqsInGroupL = it->second;
 212
 213                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupB][j], groupA, groupB);
 214                                 double weightedSum = ((numSeqsInGroupL * sum) / (double)ct->getGroupCount(groupB));
 215
 216                                 D[count] += weightedSum;
 217                         }
 218                         count++;
 219                 }
 220
 221                 //calculate u for the group comb
 222                 for (int h = start; h < (start+num); h++) {
 223                         //report progress
 224                         //m->mothurOut("Processing combo: " + toString(h)); m->mothurOutEndLine();
 225
 226                         string groupA = namesOfGroupCombos[h][0];
 227                         string groupB = namesOfGroupCombos[h][1];
 228
 229                         //calculate u for the group comb
 230                         for(int i=0;i<t->getNumNodes();i++){
 231
 232                                 if (m->control_pressed) { return data; }
 233
 234                                 double u;
 235                                 //int pcountSize = 0;
 236                                 //does this node have descendants from groupA
 237                                 it = t->tree[i].pcount.find(groupA);
 238                                 //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 239                                 if (it != t->tree[i].pcount.end()) {
 240                                         u = (double) t->tree[i].pcount[groupA] / (double) ct->getGroupCount(groupA);
 241                                 }else { u = 0.00; }
 242
 243
 244                                 //does this node have descendants from group l
 245                                 it = t->tree[i].pcount.find(groupB);
 246
 247                                 //if it does subtract their percentage from u
 248                                 if (it != t->tree[i].pcount.end()) {
 249                                         u -= (double) t->tree[i].pcount[groupB] / (double) ct->getGroupCount(groupB);
 250                                 }
 251
 252                                 if (includeRoot) {
 253                                         if (t->tree[i].getBranchLength() != -1) {
 254                                                 u = abs(u * t->tree[i].getBranchLength());
 255                                                 WScore[(groupA+groupB)] += u;
 256                                         }
 257                                 }else {
 258                                         //if this is not the root then add it
 259                                         if (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0) {
 260                                                 if (t->tree[i].getBranchLength() != -1) {
 261                                                         u = abs(u * t->tree[i].getBranchLength());
 262                                                         WScore[(groupA+groupB)] += u;
 263                                                 }
 264                                         }
 265                                 }
 266                         }
 267
 268                 }
 269
 270                 /********************************************************/
 271                 //calculate weighted score for the group combination
 272                 double UN;
 273                 count = 0;
 274                 for (int h = start; h < (start+num); h++) {
 275                         UN = (WScore[namesOfGroupCombos[h][0]+namesOfGroupCombos[h][1]] / D[count]);
 276                         if (isnan(UN) || isinf(UN)) { UN = 0; }
 277                         results.push_back(UN);
 278                         count++;
 279                 }
 280
 281                 return results;
 282         }
 283         catch(exception& e) {
 284                 m->errorOut(e, "Weighted", "driver");
 285                 exit(1);
 286         }
 287 }
 288 /**************************************************************************************************/
 289 EstOutput Weighted::getValues(Tree* t, string groupA, string groupB) {
 290  try {
 291
 292                 data.clear(); //clear out old values
 293
 294         CountTable* ct = t->getCountTable();
 295
 296                 if (m->control_pressed) { return data; }
 297
 298                 //initialize weighted score
 299                 WScore[(groupA+groupB)] = 0.0;
 300                 double D = 0.0;
 301                 set<int> validBranches;
 302
 303                 vector<string> groups; groups.push_back(groupA); groups.push_back(groupB);
 304
 305                 //adding the wieghted sums from group i
 306                 for (int j = 0; j < t->groupNodeInfo[groups[0]].size(); j++) { //the leaf nodes that have seqs from group i
 307                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[0]][j]].pcount.find(groups[0]);
 308                         int numSeqsInGroupI = it->second;
 309
 310                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[0]][j], groups[0], groups[1]);
 311                         double weightedSum = ((numSeqsInGroupI * sum) / (double)ct->getGroupCount(groups[0]));
 312
 313                         D += weightedSum;
 314                 }
 315
 316                 //adding the wieghted sums from group l
 317                 for (int j = 0; j < t->groupNodeInfo[groups[1]].size(); j++) { //the leaf nodes that have seqs from group l
 318                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[1]][j]].pcount.find(groups[1]);
 319                         int numSeqsInGroupL = it->second;
 320
 321                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[1]][j], groups[0], groups[1]);
 322                         double weightedSum = ((numSeqsInGroupL * sum) / (double)ct->getGroupCount(groups[1]));
 323
 324                         D += weightedSum;
 325                 }
 326
 327                 //calculate u for the group comb
 328                 for(int i=0;i<t->getNumNodes();i++){
 329
 330                         if (m->control_pressed) { return data; }
 331
 332                         double u;
 333                         //int pcountSize = 0;
 334                         //does this node have descendants from groupA
 335                         it = t->tree[i].pcount.find(groupA);
 336                         //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 337                         if (it != t->tree[i].pcount.end()) {
 338                                 u = (double) t->tree[i].pcount[groupA] / (double) ct->getGroupCount(groupA);
 339                         }else { u = 0.00; }
 340
 341
 342                         //does this node have descendants from group l
 343                         it = t->tree[i].pcount.find(groupB);
 344                         //if it does subtract their percentage from u
 345                         if (it != t->tree[i].pcount.end()) {
 346                                 u -= (double) t->tree[i].pcount[groupB] / (double) ct->getGroupCount(groupB);
 347                         }
 348
 349                         if (includeRoot) {
 350                                 if (t->tree[i].getBranchLength() != -1) {
 351                                         u = abs(u * t->tree[i].getBranchLength());
 352                                         WScore[(groupA+groupB)] += u;
 353                                 }
 354                         }else{
 355                                 //if this is not the root then add it
 356                                 if (rootForGrouping[groups].count(i) == 0) {
 357                                         if (t->tree[i].getBranchLength() != -1) {
 358                                                 u = abs(u * t->tree[i].getBranchLength());
 359                                                 WScore[(groupA+groupB)] += u;
 360                                         }
 361                                 }
 362                         }
 363                 }
 364                 /********************************************************/
 365
 366                 //calculate weighted score for the group combination
 367                 double UN;
 368                 UN = (WScore[(groupA+groupB)] / D);
 369
 370                 if (isnan(UN) || isinf(UN)) { UN = 0; }
 371                 data.push_back(UN);
 372
 373                 return data;
 374         }
 375         catch(exception& e) {
 376                 m->errorOut(e, "Weighted", "getValues");
 377                 exit(1);
 378         }
 379 }
 380 /**************************************************************************************************/
 381 double Weighted::getLengthToRoot(Tree* t, int v, string groupA, string groupB) {
 382         try {
 383
 384                 double sum = 0.0;
 385                 int index = v;
 386
 387                 //you are a leaf
 388                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 389                 double tempTotal = 0.0;
 390                 index = t->tree[index].getParent();
 391
 392                 vector<string> grouping; grouping.push_back(groupA); grouping.push_back(groupB);
 393
 394                 rootForGrouping[grouping].insert(index);
 395
 396                 //while you aren't at root
 397                 while(t->tree[index].getParent() != -1){
 398
 399                         if (m->control_pressed) {  return sum; }
 400
 401                         int parent = t->tree[index].getParent();
 402
 403                         if (includeRoot) { //add everyone
 404                                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 405                         }else {
 406
 407                                 //am I the root for this grouping? if so I want to stop "early"
 408                                 //does my sibling have descendants from the users groups?
 409                                 int lc = t->tree[parent].getLChild();
 410                                 int rc = t->tree[parent].getRChild();
 411
 412                                 int sib = lc;
 413                                 if (lc == index) { sib = rc; }
 414
 415                                 map<string, int>::iterator itGroup;
 416                                 int pcountSize = 0;
 417                                 itGroup = t->tree[sib].pcount.find(groupA);
 418                                 if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 419                                 itGroup = t->tree[sib].pcount.find(groupB);
 420                                 if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 421
 422                                 //if yes, I am not the root so add me
 423                                 if (pcountSize != 0) {
 424                                         if (t->tree[index].getBranchLength() != -1) {
 425                                                 sum += abs(t->tree[index].getBranchLength()) + tempTotal;
 426                                                 tempTotal = 0.0;
 427                                         }else {
 428                                                 sum += tempTotal;
 429                                                 tempTotal = 0.0;
 430                                         }
 431                                         rootForGrouping[grouping].clear();
 432                                         rootForGrouping[grouping].insert(parent);
 433                                 }else { //if no, I may be the root so add my br to tempTotal until I am proven innocent
 434                                         if (t->tree[index].getBranchLength() != -1) {
 435                                                 tempTotal += abs(t->tree[index].getBranchLength());
 436                                         }
 437                                 }
 438                         }
 439
 440                         index = parent;
 441                 }
 442
 443                 //get all nodes above the root to add so we don't add their u values above
 444                 index = *(rootForGrouping[grouping].begin());
 445
 446                 while(t->tree[index].getParent() != -1){
 447                         int parent = t->tree[index].getParent();
 448                         rootForGrouping[grouping].insert(parent);
 449                         index = parent;
 450                 }
 451                 return sum;
 452         }
 453         catch(exception& e) {
 454                 m->errorOut(e, "Weighted", "getBranchLengthSums");
 455                 exit(1);
 456         }
 457 }
 458 /**************************************************************************************************/
 459
 460