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                         count++;
 198                 }
 199
 200                 //calculate u for the group comb
 201                 for (int h = start; h < (start+num); h++) {
 202                         //report progress
 203                         //m->mothurOut("Processing combo: " + toString(h)); m->mothurOutEndLine();
 204
 205                         string groupA = namesOfGroupCombos[h][0];
 206                         string groupB = namesOfGroupCombos[h][1];
 207
 208                         //calculate u for the group comb
 209                         for(int i=0;i<t->getNumNodes();i++){
 210
 211                                 if (m->control_pressed) { return data; }
 212
 213                                 double u;
 214                                 //int pcountSize = 0;
 215                                 //does this node have descendants from groupA
 216                                 it = t->tree[i].pcount.find(groupA);
 217                                 //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 218                                 if (it != t->tree[i].pcount.end()) {
 219                                         u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
 220                                 }else { u = 0.00; }
 221
 222
 223                                 //does this node have descendants from group l
 224                                 it = t->tree[i].pcount.find(groupB);
 225
 226                                 //if it does subtract their percentage from u
 227                                 if (it != t->tree[i].pcount.end()) {
 228                                         u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 229                                 }
 230
 231                                 if (includeRoot) {
 232                                         if (t->tree[i].getBranchLength() != -1) {
 233                                                 u = abs(u * t->tree[i].getBranchLength());
 234                                                 WScore[(groupA+groupB)] += u;
 235                                         }
 236                                 }else {
 237                                         //if this is not the root then add it
 238                                         if (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0) {
 239                                                 if (t->tree[i].getBranchLength() != -1) {
 240                                                         u = abs(u * t->tree[i].getBranchLength());
 241                                                         WScore[(groupA+groupB)] += u;
 242                                                 }
 243                                         }
 244                                 }
 245                         }
 246
 247                 }
 248
 249                 /********************************************************/
 250                 //calculate weighted score for the group combination
 251                 double UN;
 252                 count = 0;
 253                 for (int h = start; h < (start+num); h++) {
 254                         UN = (WScore[namesOfGroupCombos[h][0]+namesOfGroupCombos[h][1]] / D[count]);
 255                         if (isnan(UN) || isinf(UN)) { UN = 0; }
 256                         results.push_back(UN);
 257                         count++;
 258                 }
 259
 260                 return results;
 261         }
 262         catch(exception& e) {
 263                 m->errorOut(e, "Weighted", "driver");
 264                 exit(1);
 265         }
 266 }
 267 /**************************************************************************************************/
 268 EstOutput Weighted::getValues(Tree* t, string groupA, string groupB) {
 269  try {
 270                 globaldata = GlobalData::getInstance();
 271
 272                 data.clear(); //clear out old values
 273
 274                 if (m->control_pressed) { return data; }
 275
 276                 //initialize weighted score
 277                 WScore[(groupA+groupB)] = 0.0;
 278                 double D = 0.0;
 279                 set<int> validBranches;
 280
 281                 vector<string> groups; groups.push_back(groupA); groups.push_back(groupB);
 282
 283                 //adding the wieghted sums from group i
 284                 for (int j = 0; j < t->groupNodeInfo[groups[0]].size(); j++) { //the leaf nodes that have seqs from group i
 285                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[0]][j]].pcount.find(groups[0]);
 286                         int numSeqsInGroupI = it->second;
 287
 288                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[0]][j], groups[0], groups[1]);
 289                         double weightedSum = ((numSeqsInGroupI * sum) / (double)tmap->seqsPerGroup[groups[0]]);
 290
 291                         D += weightedSum;
 292                 }
 293
 294                 //adding the wieghted sums from group l
 295                 for (int j = 0; j < t->groupNodeInfo[groups[1]].size(); j++) { //the leaf nodes that have seqs from group l
 296                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[1]][j]].pcount.find(groups[1]);
 297                         int numSeqsInGroupL = it->second;
 298
 299                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[1]][j], groups[0], groups[1]);
 300                         double weightedSum = ((numSeqsInGroupL * sum) / (double)tmap->seqsPerGroup[groups[1]]);
 301
 302                         D += weightedSum;
 303                 }
 304
 305                 //calculate u for the group comb
 306                 for(int i=0;i<t->getNumNodes();i++){
 307
 308                         if (m->control_pressed) { return data; }
 309
 310                         double u;
 311                         //int pcountSize = 0;
 312                         //does this node have descendants from groupA
 313                         it = t->tree[i].pcount.find(groupA);
 314                         //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 315                         if (it != t->tree[i].pcount.end()) {
 316                                 u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
 317                         }else { u = 0.00; }
 318
 319
 320                         //does this node have descendants from group l
 321                         it = t->tree[i].pcount.find(groupB);
 322                         //if it does subtract their percentage from u
 323                         if (it != t->tree[i].pcount.end()) {
 324                                 u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 325                         }
 326
 327                         if (includeRoot) {
 328                                 if (t->tree[i].getBranchLength() != -1) {
 329                                         u = abs(u * t->tree[i].getBranchLength());
 330                                         WScore[(groupA+groupB)] += u;
 331                                 }
 332                         }else{
 333                                 //if this is not the root then add it
 334                                 if (rootForGrouping[groups].count(i) == 0) {
 335                                         if (t->tree[i].getBranchLength() != -1) {
 336                                                 u = abs(u * t->tree[i].getBranchLength());
 337                                                 WScore[(groupA+groupB)] += u;
 338                                         }
 339                                 }
 340                         }
 341                 }
 342                 /********************************************************/
 343
 344                 //calculate weighted score for the group combination
 345                 double UN;
 346                 UN = (WScore[(groupA+groupB)] / D);
 347
 348                 if (isnan(UN) || isinf(UN)) { UN = 0; }
 349                 data.push_back(UN);
 350
 351                 return data;
 352         }
 353         catch(exception& e) {
 354                 m->errorOut(e, "Weighted", "getValues");
 355                 exit(1);
 356         }
 357 }
 358 /**************************************************************************************************/
 359 double Weighted::getLengthToRoot(Tree* t, int v, string groupA, string groupB) {
 360         try {
 361
 362                 double sum = 0.0;
 363                 int index = v;
 364
 365                 //you are a leaf
 366                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 367                 double tempTotal = 0.0;
 368                 index = t->tree[index].getParent();
 369
 370                 vector<string> grouping; grouping.push_back(groupA); grouping.push_back(groupB);
 371
 372                 rootForGrouping[grouping].insert(index);
 373
 374                 //while you aren't at root
 375                 while(t->tree[index].getParent() != -1){
 376
 377                         if (m->control_pressed) {  return sum; }
 378
 379                         int parent = t->tree[index].getParent();
 380
 381                         if (includeRoot) { //add everyone
 382                                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 383                         }else {
 384
 385                                 //am I the root for this grouping? if so I want to stop "early"
 386                                 //does my sibling have descendants from the users groups?
 387                                 int lc = t->tree[parent].getLChild();
 388                                 int rc = t->tree[parent].getRChild();
 389
 390                                 int sib = lc;
 391                                 if (lc == index) { sib = rc; }
 392
 393                                 map<string, int>::iterator itGroup;
 394                                 int pcountSize = 0;
 395                                 itGroup = t->tree[sib].pcount.find(groupA);
 396                                 if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 397                                 itGroup = t->tree[sib].pcount.find(groupB);
 398                                 if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 399
 400                                 //if yes, I am not the root so add me
 401                                 if (pcountSize != 0) {
 402                                         if (t->tree[index].getBranchLength() != -1) {
 403                                                 sum += abs(t->tree[index].getBranchLength()) + tempTotal;
 404                                                 tempTotal = 0.0;
 405                                         }else {
 406                                                 sum += tempTotal;
 407                                                 tempTotal = 0.0;
 408                                         }
 409                                         rootForGrouping[grouping].clear();
 410                                         rootForGrouping[grouping].insert(parent);
 411                                 }else { //if no, I may be the root so add my br to tempTotal until I am proven innocent
 412                                         if (t->tree[index].getBranchLength() != -1) {
 413                                                 tempTotal += abs(t->tree[index].getBranchLength());
 414                                         }
 415                                 }
 416                         }
 417
 418                         index = parent;
 419                 }
 420
 421                 //get all nodes above the root to add so we don't add their u values above
 422                 index = *(rootForGrouping[grouping].begin());
 423
 424                 while(t->tree[index].getParent() != -1){
 425                         int parent = t->tree[index].getParent();
 426                         rootForGrouping[grouping].insert(parent);
 427                         index = parent;
 428                 }
 429                 return sum;
 430         }
 431         catch(exception& e) {
 432                 m->errorOut(e, "Weighted", "getBranchLengthSums");
 433                 exit(1);
 434         }
 435 }
 436 /**************************************************************************************************/
 437
 438