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                 globaldata = GlobalData::getInstance();
  17
  18                 data.clear(); //clear out old values
  19                 int numGroups;
  20                 vector<double> D;
  21                 processors = p;
  22                 outputDir = o;
  23
  24                 numGroups = globaldata->Groups.size();
  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(globaldata->Groups[i]); groups.push_back(globaldata->Groups[l]);
  35                                 namesOfGroupCombos.push_back(groups);
  36                         }
  37                 }
  38
  39                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  40                         if(processors == 1){
  41                                 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
  42                         }else{
  43                                 int numPairs = namesOfGroupCombos.size();
  44
  45                                 int numPairsPerProcessor = numPairs / processors;
  46
  47                                 for (int i = 0; i < processors; i++) {
  48                                         int startPos = i * numPairsPerProcessor;
  49                                         if(i == processors - 1){
  50                                                 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
  51                                         }
  52                                         lines.push_back(linePair(startPos, numPairsPerProcessor));
  53                                 }
  54
  55                                 data = createProcesses(t, namesOfGroupCombos);
  56
  57                                 lines.clear();
  58                         }
  59                 #else
  60                         data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
  61                 #endif
  62
  63                 return data;
  64         }
  65         catch(exception& e) {
  66                 m->errorOut(e, "Weighted", "getValues");
  67                 exit(1);
  68         }
  69 }
  70 /**************************************************************************************************/
  71
  72 EstOutput Weighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos) {
  73         try {
  74 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  75                 int process = 1;
  76                 int num = 0;
  77                 vector<int> processIDS;
  78
  79                 EstOutput results;
  80
  81                 //loop through and create all the processes you want
  82                 while (process != processors) {
  83                         int pid = fork();
  84
  85                         if (pid > 0) {
  86                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
  87                                 process++;
  88                         }else if (pid == 0){
  89
  90                                 EstOutput Myresults;
  91                                 Myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num);
  92
  93                                 m->mothurOut("Merging results."); m->mothurOutEndLine();
  94
  95                                 //pass numSeqs to parent
  96                                 ofstream out;
  97
  98                                 string tempFile = outputDir + toString(getpid()) + ".weighted.results.temp";
  99
 100                                 m->openOutputFile(tempFile, out);
 101
 102                                 out << Myresults.size() << endl;
 103                                 for (int i = 0; i < Myresults.size(); i++) {  out << Myresults[i] << '\t';  } out << endl;
 104                                 out.close();
 105
 106                                 exit(0);
 107                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 108                 }
 109
 110                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num);
 111
 112                 //force parent to wait until all the processes are done
 113                 for (int i=0;i<(processors-1);i++) {
 114                         int temp = processIDS[i];
 115                         wait(&temp);
 116                 }
 117
 118                 if (m->control_pressed) { return results; }
 119
 120                 //get data created by processes
 121                 for (int i=0;i<(processors-1);i++) {
 122                         ifstream in;
 123                         string s = outputDir + toString(processIDS[i]) + ".weighted.results.temp";
 124                         m->openInputFile(s, in);
 125
 126                         //get quantiles
 127                         while (!in.eof()) {
 128                                 int num;
 129                                 in >> num; m->gobble(in);
 130
 131                                 if (m->control_pressed) { break; }
 132
 133                                 double w;
 134                                 for (int j = 0; j < num; j++) {
 135                                         in >> w;
 136                                         results.push_back(w);
 137                                 }
 138                                 m->gobble(in);
 139                         }
 140                         in.close();
 141                         remove(s.c_str());
 142                 }
 143
 144                 m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
 145
 146                 return results;
 147 #endif
 148         }
 149         catch(exception& e) {
 150                 m->errorOut(e, "Weighted", "createProcesses");
 151                 exit(1);
 152         }
 153 }
 154 /**************************************************************************************************/
 155 EstOutput Weighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num) {
 156  try {
 157                 EstOutput results;
 158                 vector<double> D;
 159
 160                 int count = 0;
 161                 for (int h = start; h < (start+num); h++) {
 162
 163                         if (m->control_pressed) { return results; }
 164
 165                         //initialize weighted score
 166                         string groupA = namesOfGroupCombos[h][0];
 167                         string groupB = namesOfGroupCombos[h][1];
 168
 169                         set<int> validBranches;
 170                         WScore[groupA+groupB] = 0.0;
 171                         D.push_back(0.0000); //initialize a spot in D for each combination
 172
 173                         //adding the wieghted sums from group i
 174                         for (int j = 0; j < t->groupNodeInfo[groupA].size(); j++) { //the leaf nodes that have seqs from group i
 175                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupA][j]].pcount.find(groupA);
 176                                 int numSeqsInGroupI = it->second;
 177
 178                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupA][j], groupA, groupB);
 179                                 double weightedSum = ((numSeqsInGroupI * sum) / (double)tmap->seqsPerGroup[groupA]);
 180
 181                                 D[count] += weightedSum;
 182                         }
 183
 184                         //adding the wieghted sums from group l
 185                         for (int j = 0; j < t->groupNodeInfo[groupB].size(); j++) { //the leaf nodes that have seqs from group l
 186                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupB][j]].pcount.find(groupB);
 187                                 int numSeqsInGroupL = it->second;
 188
 189                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupB][j], groupA, groupB);
 190                                 double weightedSum = ((numSeqsInGroupL * sum) / (double)tmap->seqsPerGroup[groupB]);
 191
 192                                 D[count] += weightedSum;
 193                         }
 194                         count++;
 195                 }
 196
 197                 //calculate u for the group comb
 198
 199                 for (int h = start; h < (start+num); h++) {
 200                         //report progress
 201                         m->mothurOut("Processing combo: " + toString(h)); m->mothurOutEndLine();
 202
 203                         int numLeaves = t->getNumLeaves();
 204                         map<int, double> tempTotals; //maps node to total Branch Length
 205                         map<int, int> nodePcountSize; //maps node to pcountSize
 206
 207                         string groupA = namesOfGroupCombos[h][0];
 208                         string groupB = namesOfGroupCombos[h][1];
 209
 210                         //calculate u for the group comb
 211                         for(int i=0;i<t->getNumNodes();i++){
 212
 213                                 if (m->control_pressed) { return data; }
 214
 215                                 double u;
 216                                 int pcountSize = 0;
 217                                 //does this node have descendants from groupA
 218                                 it = t->tree[i].pcount.find(groupA);
 219                                 //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 220                                 if (it != t->tree[i].pcount.end()) {
 221                                         u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
 222                                         pcountSize++;
 223                                 }else { u = 0.00; }
 224
 225
 226                                 //does this node have descendants from group l
 227                                 it = t->tree[i].pcount.find(groupB);
 228                                 //if it does subtract their percentage from u
 229                                 if (it != t->tree[i].pcount.end()) {
 230                                         u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 231                                         pcountSize++;
 232                                 }
 233
 234                                 u = abs(u * t->tree[i].getBranchLength());
 235
 236                                 nodePcountSize[i] = pcountSize;
 237
 238                                 //if you are a leaf from a users group add to total
 239                                 if (i < numLeaves) {
 240                                         if ((t->tree[i].getBranchLength() != -1) && pcountSize != 0) {
 241                                                 //cout << "added to total" << endl;
 242                                                 WScore[(groupA+groupB)] += u;
 243                                         }
 244                                         tempTotals[i] = 0.0;  //we don't care about you, or we have already added you
 245                                 }else{ //if you are not a leaf
 246                                         //do both your chidren have have descendants from the users groups?
 247                                         int lc = t->tree[i].getLChild();
 248                                         int rc = t->tree[i].getRChild();
 249
 250                                         //if yes, add your childrens tempTotals
 251                                         if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
 252                                                 WScore[(groupA+groupB)] += tempTotals[lc] + tempTotals[rc];
 253                                                 //cout << "added to total " << tempTotals[lc] << '\t' << tempTotals[rc] << endl;
 254                                                 if (t->tree[i].getBranchLength() != -1) {
 255                                                         tempTotals[i] = u;
 256                                                 }else {
 257                                                         tempTotals[i] = 0.0;
 258                                                 }
 259                                         }else if ((nodePcountSize[lc] == 0) && (nodePcountSize[rc] == 0)) { tempTotals[i] = 0.0;  //we don't care about you
 260                                         }else { //if no, your tempTotal is your childrens temp totals + your branch length
 261                                                 tempTotals[i] = tempTotals[lc] + tempTotals[rc] + u;
 262                                         }
 263                                         //cout << "temptotal = "<< tempTotals[i] << endl;
 264                                 }
 265                         }
 266                 }
 267
 268                 /********************************************************/
 269                 //calculate weighted score for the group combination
 270                 double UN;
 271                 count = 0;
 272                 for (int h = start; h < (start+num); h++) {
 273                         UN = (WScore[namesOfGroupCombos[h][0]+namesOfGroupCombos[h][1]] / D[count]);
 274
 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                 globaldata = GlobalData::getInstance();
 291
 292                 data.clear(); //clear out old values
 293
 294                 if (m->control_pressed) { return data; }
 295
 296                 //initialize weighted score
 297                 WScore[(groupA+groupB)] = 0.0;
 298                 double D = 0.0;
 299                 set<int> validBranches;
 300
 301                 vector<string> groups; groups.push_back(groupA); groups.push_back(groupB);
 302
 303                 //adding the wieghted sums from group i
 304                 for (int j = 0; j < t->groupNodeInfo[groups[0]].size(); j++) { //the leaf nodes that have seqs from group i
 305                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[0]][j]].pcount.find(groups[0]);
 306                         int numSeqsInGroupI = it->second;
 307
 308                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[0]][j], groups[0], groups[1]);
 309                         double weightedSum = ((numSeqsInGroupI * sum) / (double)tmap->seqsPerGroup[groups[0]]);
 310
 311                         D += weightedSum;
 312                 }
 313
 314                 //adding the wieghted sums from group l
 315                 for (int j = 0; j < t->groupNodeInfo[groups[1]].size(); j++) { //the leaf nodes that have seqs from group l
 316                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[1]][j]].pcount.find(groups[1]);
 317                         int numSeqsInGroupL = it->second;
 318
 319                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[1]][j], groups[0], groups[1]);
 320                         double weightedSum = ((numSeqsInGroupL * sum) / (double)tmap->seqsPerGroup[groups[1]]);
 321
 322                         D += weightedSum;
 323                 }
 324
 325                 int numLeaves = t->getNumLeaves();
 326                 map<int, double> tempTotals; //maps node to total Branch Length
 327                 map<int, int> nodePcountSize; //maps node to pcountSize
 328
 329                 //calculate u for the group comb
 330                 for(int i=0;i<t->getNumNodes();i++){
 331
 332                         if (m->control_pressed) { return data; }
 333
 334                         double u;
 335                         int pcountSize = 0;
 336                         //does this node have descendants from groupA
 337                         it = t->tree[i].pcount.find(groupA);
 338                         //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 339                         if (it != t->tree[i].pcount.end()) {
 340                                 u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
 341                                 pcountSize++;
 342                         }else { u = 0.00; }
 343
 344
 345                         //does this node have descendants from group l
 346                         it = t->tree[i].pcount.find(groupB);
 347                         //if it does subtract their percentage from u
 348                         if (it != t->tree[i].pcount.end()) {
 349                                 u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 350                                 pcountSize++;
 351                         }
 352
 353                         u = abs(u * t->tree[i].getBranchLength());
 354
 355                         nodePcountSize[i] = pcountSize;
 356
 357                         //if you are a leaf from a users group add to total
 358                         if (i < numLeaves) {
 359                                 if ((t->tree[i].getBranchLength() != -1) && pcountSize != 0) {
 360                                 //cout << "added to total" << endl;
 361                                         WScore[(groupA+groupB)] += u;
 362                                 }
 363                                 tempTotals[i] = 0.0;  //we don't care about you, or we have already added you
 364                         }else{ //if you are not a leaf
 365                                 //do both your chidren have have descendants from the users groups?
 366                                 int lc = t->tree[i].getLChild();
 367                                 int rc = t->tree[i].getRChild();
 368
 369                                 //if yes, add your childrens tempTotals
 370                                 if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
 371                                         WScore[(groupA+groupB)] += tempTotals[lc] + tempTotals[rc];
 372                                         //cout << "added to total " << tempTotals[lc] << '\t' << tempTotals[rc] << endl;
 373                                         if (t->tree[i].getBranchLength() != -1) {
 374                                                 tempTotals[i] = u;
 375                                         }else {
 376                                                 tempTotals[i] = 0.0;
 377                                         }
 378                                 }else if ((nodePcountSize[lc] == 0) && (nodePcountSize[rc] == 0)) { tempTotals[i] = 0.0;  //we don't care about you
 379                                 }else { //if no, your tempTotal is your childrens temp totals + your branch length
 380                                         tempTotals[i] = tempTotals[lc] + tempTotals[rc] + u;
 381                                 }
 382                                 //cout << "temptotal = "<< tempTotals[i] << endl;
 383                         }
 384                 }
 385
 386                 /********************************************************/
 387
 388                 //calculate weighted score for the group combination
 389                 double UN;
 390                 UN = (WScore[(groupA+groupB)] / D);
 391
 392                 if (isnan(UN) || isinf(UN)) { UN = 0; }
 393                 data.push_back(UN);
 394
 395                 return data;
 396         }
 397         catch(exception& e) {
 398                 m->errorOut(e, "Weighted", "getValues");
 399                 exit(1);
 400         }
 401 }
 402 /**************************************************************************************************/
 403 double Weighted::getLengthToRoot(Tree* t, int v, string groupA, string groupB) {
 404         try {
 405
 406                 double sum = 0.0;
 407                 map<int, double> tempTotals; //maps node to total Branch Length
 408                 map<int, int> nodePcountSize; //maps node to pcountSize
 409                 map<int, int>::iterator itCount;
 410
 411                 int index = v;
 412
 413                 //you are a leaf
 414                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 415                 tempTotals[index] = 0.0;
 416                 index = t->tree[index].getParent();
 417
 418                 //while you aren't at root
 419                 while(t->tree[index].getParent() != -1){
 420
 421                         if (m->control_pressed) {  return sum; }
 422
 423                         int pcountSize = 0;
 424                         map<string, int>::iterator itGroup = t->tree[index].pcount.find(groupA);
 425                         if (itGroup != t->tree[index].pcount.end()) { pcountSize++;  }
 426                         itGroup = t->tree[index].pcount.find(groupB);
 427                         if (itGroup != t->tree[index].pcount.end()) { pcountSize++;  }
 428
 429                         nodePcountSize[index] = pcountSize;
 430
 431                         //do both your chidren have have descendants from the users groups?
 432                         int lc = t->tree[index].getLChild();
 433                         int rc = t->tree[index].getRChild();
 434
 435                         itCount = nodePcountSize.find(lc);
 436                         if (itCount == nodePcountSize.end()) {
 437                                 int LpcountSize = 0;
 438                                 itGroup = t->tree[lc].pcount.find(groupA);
 439                                 if (itGroup != t->tree[lc].pcount.end()) { LpcountSize++;  }
 440                                 itGroup = t->tree[lc].pcount.find(groupB);
 441                                 if (itGroup != t->tree[lc].pcount.end()) { LpcountSize++;  }
 442                                 nodePcountSize[lc] = LpcountSize;
 443                         }
 444
 445                         itCount = nodePcountSize.find(rc);
 446                         if (itCount == nodePcountSize.end()) {
 447                                 int RpcountSize = 0;
 448                                 itGroup = t->tree[rc].pcount.find(groupA);
 449                                 if (itGroup != t->tree[rc].pcount.end()) { RpcountSize++;  }
 450                                 itGroup = t->tree[rc].pcount.find(groupB);
 451                                 if (itGroup != t->tree[rc].pcount.end()) { RpcountSize++;  }
 452                                 nodePcountSize[rc] = RpcountSize;
 453                         }
 454
 455                         //if yes, add your childrens tempTotals
 456                         if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
 457                                 sum += tempTotals[lc] + tempTotals[rc];
 458
 459                                 //cout << "added to total " << tempTotals[lc] << '\t' << tempTotals[rc] << endl;
 460                                 if (t->tree[index].getBranchLength() != -1) {
 461                                         tempTotals[index] = abs(t->tree[index].getBranchLength());
 462                                 }else {
 463                                         tempTotals[index] = 0.0;
 464                                 }
 465                         }else { //if no, your tempTotal is your childrens temp totals + your branch length
 466                                 tempTotals[index] = tempTotals[lc] + tempTotals[rc] + abs(t->tree[index].getBranchLength());
 467                         }
 468                         //cout << "temptotal = "<< tempTotals[i] << endl;
 469
 470                         index = t->tree[index].getParent();
 471                 }
 472
 473                 return sum;
 474         }
 475         catch(exception& e) {
 476                 m->errorOut(e, "Weighted", "getBranchLengthSums");
 477                 exit(1);
 478         }
 479 }
 480 /**************************************************************************************************/
 481
 482
 483