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 {
 108                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 109                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 110                                 exit(0);
 111                         }
 112                 }
 113
 114                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num);
 115
 116                 //force parent to wait until all the processes are done
 117                 for (int i=0;i<(processors-1);i++) {
 118                         int temp = processIDS[i];
 119                         wait(&temp);
 120                 }
 121
 122                 if (m->control_pressed) { return results; }
 123
 124                 //get data created by processes
 125                 for (int i=0;i<(processors-1);i++) {
 126                         ifstream in;
 127                         string s = outputDir + toString(processIDS[i]) + ".weighted.results.temp";
 128                         m->openInputFile(s, in);
 129
 130                         //get quantiles
 131                         while (!in.eof()) {
 132                                 int num;
 133                                 in >> num; m->gobble(in);
 134
 135                                 if (m->control_pressed) { break; }
 136
 137                                 double w;
 138                                 for (int j = 0; j < num; j++) {
 139                                         in >> w;
 140                                         results.push_back(w);
 141                                 }
 142                                 m->gobble(in);
 143                         }
 144                         in.close();
 145                         remove(s.c_str());
 146                 }
 147
 148                 m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
 149
 150                 return results;
 151 #endif
 152         }
 153         catch(exception& e) {
 154                 m->errorOut(e, "Weighted", "createProcesses");
 155                 exit(1);
 156         }
 157 }
 158 /**************************************************************************************************/
 159 EstOutput Weighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num) {
 160  try {
 161                 EstOutput results;
 162                 vector<double> D;
 163
 164                 int count = 0;
 165                 for (int h = start; h < (start+num); h++) {
 166
 167                         if (m->control_pressed) { return results; }
 168
 169                         //initialize weighted score
 170                         string groupA = namesOfGroupCombos[h][0];
 171                         string groupB = namesOfGroupCombos[h][1];
 172
 173                         set<int> validBranches;
 174                         WScore[groupA+groupB] = 0.0;
 175                         D.push_back(0.0000); //initialize a spot in D for each combination
 176
 177                         //adding the wieghted sums from group i
 178                         for (int j = 0; j < t->groupNodeInfo[groupA].size(); j++) { //the leaf nodes that have seqs from group i
 179                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupA][j]].pcount.find(groupA);
 180                                 int numSeqsInGroupI = it->second;
 181
 182                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupA][j], groupA, groupB);
 183                                 double weightedSum = ((numSeqsInGroupI * sum) / (double)tmap->seqsPerGroup[groupA]);
 184
 185                                 D[count] += weightedSum;
 186                         }
 187
 188                         //adding the wieghted sums from group l
 189                         for (int j = 0; j < t->groupNodeInfo[groupB].size(); j++) { //the leaf nodes that have seqs from group l
 190                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupB][j]].pcount.find(groupB);
 191                                 int numSeqsInGroupL = it->second;
 192
 193                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupB][j], groupA, groupB);
 194                                 double weightedSum = ((numSeqsInGroupL * sum) / (double)tmap->seqsPerGroup[groupB]);
 195
 196                                 D[count] += weightedSum;
 197                         }
 198                         count++;
 199                 }
 200
 201                 //calculate u for the group comb
 202
 203                 for (int h = start; h < (start+num); h++) {
 204                         //report progress
 205                         m->mothurOut("Processing combo: " + toString(h)); m->mothurOutEndLine();
 206
 207                         int numLeaves = t->getNumLeaves();
 208                         map<int, double> tempTotals; //maps node to total Branch Length
 209                         map<int, int> nodePcountSize; //maps node to pcountSize
 210
 211                         string groupA = namesOfGroupCombos[h][0];
 212                         string groupB = namesOfGroupCombos[h][1];
 213
 214                         //calculate u for the group comb
 215                         for(int i=0;i<t->getNumNodes();i++){
 216
 217                                 if (m->control_pressed) { return data; }
 218
 219                                 double u;
 220                                 int pcountSize = 0;
 221                                 //does this node have descendants from groupA
 222                                 it = t->tree[i].pcount.find(groupA);
 223                                 //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 224                                 if (it != t->tree[i].pcount.end()) {
 225                                         u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
 226                                         pcountSize++;
 227                                 }else { u = 0.00; }
 228
 229
 230                                 //does this node have descendants from group l
 231                                 it = t->tree[i].pcount.find(groupB);
 232                                 //if it does subtract their percentage from u
 233                                 if (it != t->tree[i].pcount.end()) {
 234                                         u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 235                                         pcountSize++;
 236                                 }
 237
 238                                 u = abs(u * t->tree[i].getBranchLength());
 239
 240                                 nodePcountSize[i] = pcountSize;
 241
 242                                 //if you are a leaf from a users group add to total
 243                                 if (i < numLeaves) {
 244                                         if ((t->tree[i].getBranchLength() != -1) && pcountSize != 0) {
 245                                                 //cout << "added to total" << endl;
 246                                                 WScore[(groupA+groupB)] += u;
 247                                         }
 248                                         tempTotals[i] = 0.0;  //we don't care about you, or we have already added you
 249                                 }else{ //if you are not a leaf
 250                                         //do both your chidren have have descendants from the users groups?
 251                                         int lc = t->tree[i].getLChild();
 252                                         int rc = t->tree[i].getRChild();
 253
 254                                         //if yes, add your childrens tempTotals
 255                                         if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
 256                                                 WScore[(groupA+groupB)] += tempTotals[lc] + tempTotals[rc];
 257                                                 //cout << "added to total " << tempTotals[lc] << '\t' << tempTotals[rc] << endl;
 258                                                 if (t->tree[i].getBranchLength() != -1) {
 259                                                         tempTotals[i] = u;
 260                                                 }else {
 261                                                         tempTotals[i] = 0.0;
 262                                                 }
 263                                         }else if ((nodePcountSize[lc] == 0) && (nodePcountSize[rc] == 0)) { tempTotals[i] = 0.0;  //we don't care about you
 264                                         }else { //if no, your tempTotal is your childrens temp totals + your branch length
 265                                                 tempTotals[i] = tempTotals[lc] + tempTotals[rc] + u;
 266                                         }
 267                                         //cout << "temptotal = "<< tempTotals[i] << endl;
 268                                 }
 269                         }
 270                 }
 271
 272                 /********************************************************/
 273                 //calculate weighted score for the group combination
 274                 double UN;
 275                 count = 0;
 276                 for (int h = start; h < (start+num); h++) {
 277                         UN = (WScore[namesOfGroupCombos[h][0]+namesOfGroupCombos[h][1]] / D[count]);
 278
 279                         if (isnan(UN) || isinf(UN)) { UN = 0; }
 280                         results.push_back(UN);
 281                         count++;
 282                 }
 283
 284                 return results;
 285         }
 286         catch(exception& e) {
 287                 m->errorOut(e, "Weighted", "driver");
 288                 exit(1);
 289         }
 290 }
 291 /**************************************************************************************************/
 292 EstOutput Weighted::getValues(Tree* t, string groupA, string groupB) {
 293  try {
 294                 globaldata = GlobalData::getInstance();
 295
 296                 data.clear(); //clear out old values
 297
 298                 if (m->control_pressed) { return data; }
 299
 300                 //initialize weighted score
 301                 WScore[(groupA+groupB)] = 0.0;
 302                 double D = 0.0;
 303                 set<int> validBranches;
 304
 305                 vector<string> groups; groups.push_back(groupA); groups.push_back(groupB);
 306
 307                 //adding the wieghted sums from group i
 308                 for (int j = 0; j < t->groupNodeInfo[groups[0]].size(); j++) { //the leaf nodes that have seqs from group i
 309                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[0]][j]].pcount.find(groups[0]);
 310                         int numSeqsInGroupI = it->second;
 311
 312                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[0]][j], groups[0], groups[1]);
 313                         double weightedSum = ((numSeqsInGroupI * sum) / (double)tmap->seqsPerGroup[groups[0]]);
 314
 315                         D += weightedSum;
 316                 }
 317
 318                 //adding the wieghted sums from group l
 319                 for (int j = 0; j < t->groupNodeInfo[groups[1]].size(); j++) { //the leaf nodes that have seqs from group l
 320                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[1]][j]].pcount.find(groups[1]);
 321                         int numSeqsInGroupL = it->second;
 322
 323                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[1]][j], groups[0], groups[1]);
 324                         double weightedSum = ((numSeqsInGroupL * sum) / (double)tmap->seqsPerGroup[groups[1]]);
 325
 326                         D += weightedSum;
 327                 }
 328
 329                 int numLeaves = t->getNumLeaves();
 330                 map<int, double> tempTotals; //maps node to total Branch Length
 331                 map<int, int> nodePcountSize; //maps node to pcountSize
 332
 333                 //calculate u for the group comb
 334                 for(int i=0;i<t->getNumNodes();i++){
 335
 336                         if (m->control_pressed) { return data; }
 337
 338                         double u;
 339                         int pcountSize = 0;
 340                         //does this node have descendants from groupA
 341                         it = t->tree[i].pcount.find(groupA);
 342                         //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 343                         if (it != t->tree[i].pcount.end()) {
 344                                 u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
 345                                 pcountSize++;
 346                         }else { u = 0.00; }
 347
 348
 349                         //does this node have descendants from group l
 350                         it = t->tree[i].pcount.find(groupB);
 351                         //if it does subtract their percentage from u
 352                         if (it != t->tree[i].pcount.end()) {
 353                                 u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 354                                 pcountSize++;
 355                         }
 356
 357                         u = abs(u * t->tree[i].getBranchLength());
 358
 359                         nodePcountSize[i] = pcountSize;
 360
 361                         //if you are a leaf from a users group add to total
 362                         if (i < numLeaves) {
 363                                 if ((t->tree[i].getBranchLength() != -1) && pcountSize != 0) {
 364                                 //cout << "added to total" << endl;
 365                                         WScore[(groupA+groupB)] += u;
 366                                 }
 367                                 tempTotals[i] = 0.0;  //we don't care about you, or we have already added you
 368                         }else{ //if you are not a leaf
 369                                 //do both your chidren have have descendants from the users groups?
 370                                 int lc = t->tree[i].getLChild();
 371                                 int rc = t->tree[i].getRChild();
 372
 373                                 //if yes, add your childrens tempTotals
 374                                 if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
 375                                         WScore[(groupA+groupB)] += tempTotals[lc] + tempTotals[rc];
 376                                         //cout << "added to total " << tempTotals[lc] << '\t' << tempTotals[rc] << endl;
 377                                         if (t->tree[i].getBranchLength() != -1) {
 378                                                 tempTotals[i] = u;
 379                                         }else {
 380                                                 tempTotals[i] = 0.0;
 381                                         }
 382                                 }else if ((nodePcountSize[lc] == 0) && (nodePcountSize[rc] == 0)) { tempTotals[i] = 0.0;  //we don't care about you
 383                                 }else { //if no, your tempTotal is your childrens temp totals + your branch length
 384                                         tempTotals[i] = tempTotals[lc] + tempTotals[rc] + u;
 385                                 }
 386                                 //cout << "temptotal = "<< tempTotals[i] << endl;
 387                         }
 388                 }
 389
 390                 /********************************************************/
 391
 392                 //calculate weighted score for the group combination
 393                 double UN;
 394                 UN = (WScore[(groupA+groupB)] / D);
 395
 396                 if (isnan(UN) || isinf(UN)) { UN = 0; }
 397                 data.push_back(UN);
 398
 399                 return data;
 400         }
 401         catch(exception& e) {
 402                 m->errorOut(e, "Weighted", "getValues");
 403                 exit(1);
 404         }
 405 }
 406 /**************************************************************************************************/
 407 double Weighted::getLengthToRoot(Tree* t, int v, string groupA, string groupB) {
 408         try {
 409
 410                 double sum = 0.0;
 411                 map<int, double> tempTotals; //maps node to total Branch Length
 412                 map<int, int> nodePcountSize; //maps node to pcountSize
 413                 map<int, int>::iterator itCount;
 414
 415                 int index = v;
 416
 417                 //you are a leaf
 418                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 419                 tempTotals[index] = 0.0;
 420                 index = t->tree[index].getParent();
 421
 422                 //while you aren't at root
 423                 while(t->tree[index].getParent() != -1){
 424
 425                         if (m->control_pressed) {  return sum; }
 426
 427                         int pcountSize = 0;
 428                         map<string, int>::iterator itGroup = t->tree[index].pcount.find(groupA);
 429                         if (itGroup != t->tree[index].pcount.end()) { pcountSize++;  }
 430                         itGroup = t->tree[index].pcount.find(groupB);
 431                         if (itGroup != t->tree[index].pcount.end()) { pcountSize++;  }
 432
 433                         nodePcountSize[index] = pcountSize;
 434
 435                         //do both your chidren have have descendants from the users groups?
 436                         int lc = t->tree[index].getLChild();
 437                         int rc = t->tree[index].getRChild();
 438
 439                         itCount = nodePcountSize.find(lc);
 440                         if (itCount == nodePcountSize.end()) {
 441                                 int LpcountSize = 0;
 442                                 itGroup = t->tree[lc].pcount.find(groupA);
 443                                 if (itGroup != t->tree[lc].pcount.end()) { LpcountSize++;  }
 444                                 itGroup = t->tree[lc].pcount.find(groupB);
 445                                 if (itGroup != t->tree[lc].pcount.end()) { LpcountSize++;  }
 446                                 nodePcountSize[lc] = LpcountSize;
 447                         }
 448
 449                         itCount = nodePcountSize.find(rc);
 450                         if (itCount == nodePcountSize.end()) {
 451                                 int RpcountSize = 0;
 452                                 itGroup = t->tree[rc].pcount.find(groupA);
 453                                 if (itGroup != t->tree[rc].pcount.end()) { RpcountSize++;  }
 454                                 itGroup = t->tree[rc].pcount.find(groupB);
 455                                 if (itGroup != t->tree[rc].pcount.end()) { RpcountSize++;  }
 456                                 nodePcountSize[rc] = RpcountSize;
 457                         }
 458
 459                         //if yes, add your childrens tempTotals
 460                         if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
 461                                 sum += tempTotals[lc] + tempTotals[rc];
 462
 463                                 //cout << "added to total " << tempTotals[lc] << '\t' << tempTotals[rc] << endl;
 464                                 if (t->tree[index].getBranchLength() != -1) {
 465                                         tempTotals[index] = abs(t->tree[index].getBranchLength());
 466                                 }else {
 467                                         tempTotals[index] = 0.0;
 468                                 }
 469                         }else { //if no, your tempTotal is your childrens temp totals + your branch length
 470                                 tempTotals[index] = tempTotals[lc] + tempTotals[rc] + abs(t->tree[index].getBranchLength());
 471                         }
 472                         //cout << "temptotal = "<< tempTotals[i] << endl;
 473
 474                         index = t->tree[index].getParent();
 475                 }
 476
 477                 return sum;
 478         }
 479         catch(exception& e) {
 480                 m->errorOut(e, "Weighted", "getBranchLengthSums");
 481                 exit(1);
 482         }
 483 }
 484 /**************************************************************************************************/
 485
 486
 487