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                 vector<int> processIDS;
  77
  78                 EstOutput results;
  79
  80                 //loop through and create all the processes you want
  81                 while (process != processors) {
  82                         int pid = fork();
  83
  84                         if (pid > 0) {
  85                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
  86                                 process++;
  87                         }else if (pid == 0){
  88
  89                                 EstOutput Myresults;
  90                                 Myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num);
  91
  92                                 m->mothurOut("Merging results."); m->mothurOutEndLine();
  93
  94                                 //pass numSeqs to parent
  95                                 ofstream out;
  96
  97                                 string tempFile = outputDir + toString(getpid()) + ".weighted.results.temp";
  98
  99                                 m->openOutputFile(tempFile, out);
 100
 101                                 out << Myresults.size() << endl;
 102                                 for (int i = 0; i < Myresults.size(); i++) {  out << Myresults[i] << '\t';  } out << endl;
 103                                 out.close();
 104
 105                                 exit(0);
 106                         }else {
 107                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 108                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 109                                 exit(0);
 110                         }
 111                 }
 112
 113                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num);
 114
 115                 //force parent to wait until all the processes are done
 116                 for (int i=0;i<(processors-1);i++) {
 117                         int temp = processIDS[i];
 118                         wait(&temp);
 119                 }
 120
 121                 if (m->control_pressed) { return results; }
 122
 123                 //get data created by processes
 124                 for (int i=0;i<(processors-1);i++) {
 125                         ifstream in;
 126                         string s = outputDir + toString(processIDS[i]) + ".weighted.results.temp";
 127                         m->openInputFile(s, in);
 128
 129                         //get quantiles
 130                         while (!in.eof()) {
 131                                 int num;
 132                                 in >> num; m->gobble(in);
 133
 134                                 if (m->control_pressed) { break; }
 135
 136                                 double w;
 137                                 for (int j = 0; j < num; j++) {
 138                                         in >> w;
 139                                         results.push_back(w);
 140                                 }
 141                                 m->gobble(in);
 142                         }
 143                         in.close();
 144                         remove(s.c_str());
 145                 }
 146
 147                 m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
 148
 149                 return results;
 150 #endif
 151         }
 152         catch(exception& e) {
 153                 m->errorOut(e, "Weighted", "createProcesses");
 154                 exit(1);
 155         }
 156 }
 157 /**************************************************************************************************/
 158 EstOutput Weighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num) {
 159  try {
 160                 EstOutput results;
 161                 vector<double> D;
 162
 163                 int count = 0;
 164                 for (int h = start; h < (start+num); h++) {
 165
 166                         if (m->control_pressed) { return results; }
 167
 168                         //initialize weighted score
 169                         string groupA = namesOfGroupCombos[h][0];
 170                         string groupB = namesOfGroupCombos[h][1];
 171
 172                         set<int> validBranches;
 173                         WScore[groupA+groupB] = 0.0;
 174                         D.push_back(0.0000); //initialize a spot in D for each combination
 175
 176                         //adding the wieghted sums from group i
 177                         for (int j = 0; j < t->groupNodeInfo[groupA].size(); j++) { //the leaf nodes that have seqs from group i
 178                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupA][j]].pcount.find(groupA);
 179                                 int numSeqsInGroupI = it->second;
 180
 181                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupA][j], groupA, groupB);
 182                                 double weightedSum = ((numSeqsInGroupI * sum) / (double)tmap->seqsPerGroup[groupA]);
 183
 184                                 D[count] += weightedSum;
 185                         }
 186
 187                         //adding the wieghted sums from group l
 188                         for (int j = 0; j < t->groupNodeInfo[groupB].size(); j++) { //the leaf nodes that have seqs from group l
 189                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupB][j]].pcount.find(groupB);
 190                                 int numSeqsInGroupL = it->second;
 191
 192                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupB][j], groupA, groupB);
 193                                 double weightedSum = ((numSeqsInGroupL * sum) / (double)tmap->seqsPerGroup[groupB]);
 194
 195                                 D[count] += weightedSum;
 196                         }
 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                         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                                 }else { u = 0.00; }
 223
 224
 225                                 //does this node have descendants from group l
 226                                 it = t->tree[i].pcount.find(groupB);
 227                                 //if it does subtract their percentage from u
 228                                 if (it != t->tree[i].pcount.end()) {
 229                                         u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 230                                 }
 231
 232                                 //if this is not the root then add it
 233                                 if (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0) {
 234                                         if (t->tree[i].getBranchLength() != -1) {
 235                                                 u = abs(u * t->tree[i].getBranchLength());
 236                                                 WScore[(groupA+groupB)] += u;
 237                                         }
 238                                 }
 239
 240
 241                         }
 242                 }
 243
 244                 /********************************************************/
 245                 //calculate weighted score for the group combination
 246                 double UN;
 247                 count = 0;
 248                 for (int h = start; h < (start+num); h++) {
 249                         UN = (WScore[namesOfGroupCombos[h][0]+namesOfGroupCombos[h][1]] / D[count]);
 250                         if (isnan(UN) || isinf(UN)) { UN = 0; }
 251                         results.push_back(UN);
 252                         count++;
 253                 }
 254
 255                 return results;
 256         }
 257         catch(exception& e) {
 258                 m->errorOut(e, "Weighted", "driver");
 259                 exit(1);
 260         }
 261 }
 262 /**************************************************************************************************/
 263 EstOutput Weighted::getValues(Tree* t, string groupA, string groupB) {
 264  try {
 265                 globaldata = GlobalData::getInstance();
 266
 267                 data.clear(); //clear out old values
 268
 269                 if (m->control_pressed) { return data; }
 270
 271                 //initialize weighted score
 272                 WScore[(groupA+groupB)] = 0.0;
 273                 double D = 0.0;
 274                 set<int> validBranches;
 275
 276                 vector<string> groups; groups.push_back(groupA); groups.push_back(groupB);
 277
 278                 //adding the wieghted sums from group i
 279                 for (int j = 0; j < t->groupNodeInfo[groups[0]].size(); j++) { //the leaf nodes that have seqs from group i
 280                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[0]][j]].pcount.find(groups[0]);
 281                         int numSeqsInGroupI = it->second;
 282
 283                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[0]][j], groups[0], groups[1]);
 284                         double weightedSum = ((numSeqsInGroupI * sum) / (double)tmap->seqsPerGroup[groups[0]]);
 285
 286                         D += weightedSum;
 287                 }
 288
 289                 //adding the wieghted sums from group l
 290                 for (int j = 0; j < t->groupNodeInfo[groups[1]].size(); j++) { //the leaf nodes that have seqs from group l
 291                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[1]][j]].pcount.find(groups[1]);
 292                         int numSeqsInGroupL = it->second;
 293
 294                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[1]][j], groups[0], groups[1]);
 295                         double weightedSum = ((numSeqsInGroupL * sum) / (double)tmap->seqsPerGroup[groups[1]]);
 296
 297                         D += weightedSum;
 298                 }
 299
 300                 //calculate u for the group comb
 301                 for(int i=0;i<t->getNumNodes();i++){
 302
 303                         if (m->control_pressed) { return data; }
 304
 305                         double u;
 306                         //int pcountSize = 0;
 307                         //does this node have descendants from groupA
 308                         it = t->tree[i].pcount.find(groupA);
 309                         //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 310                         if (it != t->tree[i].pcount.end()) {
 311                                 u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
 312                         }else { u = 0.00; }
 313
 314
 315                         //does this node have descendants from group l
 316                         it = t->tree[i].pcount.find(groupB);
 317                         //if it does subtract their percentage from u
 318                         if (it != t->tree[i].pcount.end()) {
 319                                 u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 320                         }
 321
 322                         //if this is not the root then add it
 323                         if (rootForGrouping[groups].count(i) == 0) {
 324                                 if (t->tree[i].getBranchLength() != -1) {
 325                                         u = abs(u * t->tree[i].getBranchLength());
 326                                         WScore[(groupA+groupB)] += u;
 327                                 }
 328                         }
 329                 }
 330                 /********************************************************/
 331
 332                 //calculate weighted score for the group combination
 333                 double UN;
 334                 UN = (WScore[(groupA+groupB)] / D);
 335
 336                 if (isnan(UN) || isinf(UN)) { UN = 0; }
 337                 data.push_back(UN);
 338
 339                 return data;
 340         }
 341         catch(exception& e) {
 342                 m->errorOut(e, "Weighted", "getValues");
 343                 exit(1);
 344         }
 345 }
 346 /**************************************************************************************************/
 347 double Weighted::getLengthToRoot(Tree* t, int v, string groupA, string groupB) {
 348         try {
 349
 350                 double sum = 0.0;
 351                 int index = v;
 352
 353                 //you are a leaf
 354                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 355                 double tempTotal = 0.0;
 356                 index = t->tree[index].getParent();
 357
 358                 vector<string> grouping; grouping.push_back(groupA); grouping.push_back(groupB);
 359
 360                 rootForGrouping[grouping].insert(index);
 361
 362                 //while you aren't at root
 363                 while(t->tree[index].getParent() != -1){
 364
 365                         if (m->control_pressed) {  return sum; }
 366
 367                         //am I the root for this grouping? if so I want to stop "early"
 368                         //does my sibling have descendants from the users groups?
 369                         int parent = t->tree[index].getParent();
 370                         int lc = t->tree[parent].getLChild();
 371                         int rc = t->tree[parent].getRChild();
 372
 373                         int sib = lc;
 374                         if (lc == index) { sib = rc; }
 375
 376                         map<string, int>::iterator itGroup;
 377                         int pcountSize = 0;
 378                         itGroup = t->tree[sib].pcount.find(groupA);
 379                         if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 380                         itGroup = t->tree[sib].pcount.find(groupB);
 381                         if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 382
 383                         //if yes, I am not the root so add me
 384                         if (pcountSize != 0) {
 385                                 if (t->tree[index].getBranchLength() != -1) {
 386                                         sum += abs(t->tree[index].getBranchLength()) + tempTotal;
 387                                         tempTotal = 0.0;
 388                                 }else {
 389                                         sum += tempTotal;
 390                                         tempTotal = 0.0;
 391                                 }
 392                                 rootForGrouping[grouping].clear();
 393                                 rootForGrouping[grouping].insert(parent);
 394                         }else { //if no, I may be the root so add my br to tempTotal until I am proven innocent
 395                                 if (t->tree[index].getBranchLength() != -1) {
 396                                         tempTotal += abs(t->tree[index].getBranchLength());
 397                                 }
 398                         }
 399
 400                         index = parent;
 401                 }
 402
 403                 //get all nodes above the root to add so we don't add their u values above
 404                 index = *(rootForGrouping[grouping].begin());
 405                 while(t->tree[index].getParent() != -1){
 406                         int parent = t->tree[index].getParent();
 407                         rootForGrouping[grouping].insert(parent);
 408                         index = parent;
 409                 }
 410
 411                 return sum;
 412         }
 413         catch(exception& e) {
 414                 m->errorOut(e, "Weighted", "getBranchLengthSums");
 415                 exit(1);
 416         }
 417 }
 418 /**************************************************************************************************/
 419
 420