weighted.cpp

   1 /*
   2  *  weighted.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 2/9/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "weighted.h"
  11
  12 /**************************************************************************************************/
  13
  14 EstOutput Weighted::getValues(Tree* t, int p, string o) {
  15     try {
  16                 data.clear(); //clear out old values
  17                 int numGroups;
  18                 vector<double> D;
  19                 processors = p;
  20                 outputDir = o;
  21
  22                 numGroups = m->getNumGroups();
  23
  24                 if (m->control_pressed) { return data; }
  25
  26                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
  27                 vector< vector<string> > namesOfGroupCombos;
  28                 for (int i=0; i<numGroups; i++) {
  29                         for (int l = 0; l < i; l++) {
  30                                 //initialize weighted scores
  31                                 //WScore[globaldata->Groups[i]+globaldata->Groups[l]] = 0.0;
  32                                 vector<string> groups; groups.push_back((m->getGroups())[i]); groups.push_back((m->getGroups())[l]);
  33                                 namesOfGroupCombos.push_back(groups);
  34                         }
  35                 }
  36
  37                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  38                         if(processors == 1){
  39                                 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
  40                         }else{
  41                                 int numPairs = namesOfGroupCombos.size();
  42
  43                                 int numPairsPerProcessor = numPairs / processors;
  44
  45                                 for (int i = 0; i < processors; i++) {
  46                                         int startPos = i * numPairsPerProcessor;
  47                                         if(i == processors - 1){
  48                                                 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
  49                                         }
  50                                         lines.push_back(linePair(startPos, numPairsPerProcessor));
  51                                 }
  52
  53                                 data = createProcesses(t, namesOfGroupCombos);
  54
  55                                 lines.clear();
  56                         }
  57                 #else
  58                         data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
  59                 #endif
  60
  61                 return data;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "Weighted", "getValues");
  65                 exit(1);
  66         }
  67 }
  68 /**************************************************************************************************/
  69
  70 EstOutput Weighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos) {
  71         try {
  72 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  73                 int process = 1;
  74                 vector<int> processIDS;
  75
  76                 EstOutput results;
  77
  78                 //loop through and create all the processes you want
  79                 while (process != processors) {
  80                         int pid = fork();
  81
  82                         if (pid > 0) {
  83                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
  84                                 process++;
  85                         }else if (pid == 0){
  86
  87                                 EstOutput Myresults;
  88                                 Myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num);
  89
  90                                 m->mothurOut("Merging results."); m->mothurOutEndLine();
  91
  92                                 //pass numSeqs to parent
  93                                 ofstream out;
  94
  95                                 string tempFile = outputDir + toString(getpid()) + ".weighted.results.temp";
  96
  97                                 m->openOutputFile(tempFile, out);
  98
  99                                 out << Myresults.size() << endl;
 100                                 for (int i = 0; i < Myresults.size(); i++) {  out << Myresults[i] << '\t';  } out << endl;
 101                                 out.close();
 102
 103                                 exit(0);
 104                         }else {
 105                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 106                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 107                                 exit(0);
 108                         }
 109                 }
 110
 111                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num);
 112
 113                 //force parent to wait until all the processes are done
 114                 for (int i=0;i<(processors-1);i++) {
 115                         int temp = processIDS[i];
 116                         wait(&temp);
 117                 }
 118
 119                 if (m->control_pressed) { return results; }
 120
 121                 //get data created by processes
 122                 for (int i=0;i<(processors-1);i++) {
 123                         ifstream in;
 124                         string s = outputDir + toString(processIDS[i]) + ".weighted.results.temp";
 125                         m->openInputFile(s, in);
 126
 127                         //get quantiles
 128                         while (!in.eof()) {
 129                                 int num;
 130                                 in >> num; m->gobble(in);
 131
 132                                 if (m->control_pressed) { break; }
 133
 134                                 double w;
 135                                 for (int j = 0; j < num; j++) {
 136                                         in >> w;
 137                                         results.push_back(w);
 138                                 }
 139                                 m->gobble(in);
 140                         }
 141                         in.close();
 142                         m->mothurRemove(s);
 143                 }
 144
 145                 m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
 146
 147                 return results;
 148 #endif
 149         }
 150         catch(exception& e) {
 151                 m->errorOut(e, "Weighted", "createProcesses");
 152                 exit(1);
 153         }
 154 }
 155 /**************************************************************************************************/
 156 EstOutput Weighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num) {
 157  try {
 158                 EstOutput results;
 159                 vector<double> D;
 160
 161                 int count = 0;
 162                 for (int h = start; h < (start+num); h++) {
 163
 164                         if (m->control_pressed) { return results; }
 165
 166                         //initialize weighted score
 167                         string groupA = namesOfGroupCombos[h][0];
 168                         string groupB = namesOfGroupCombos[h][1];
 169
 170                         set<int> validBranches;
 171                         WScore[groupA+groupB] = 0.0;
 172                         D.push_back(0.0000); //initialize a spot in D for each combination
 173
 174                         //adding the wieghted sums from group i
 175                         for (int j = 0; j < t->groupNodeInfo[groupA].size(); j++) { //the leaf nodes that have seqs from group i
 176                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupA][j]].pcount.find(groupA);
 177                                 int numSeqsInGroupI = it->second;
 178
 179                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupA][j], groupA, groupB);
 180                                 double weightedSum = ((numSeqsInGroupI * sum) / (double)tmap->seqsPerGroup[groupA]);
 181
 182                                 D[count] += weightedSum;
 183                         }
 184
 185                         //adding the wieghted sums from group l
 186                         for (int j = 0; j < t->groupNodeInfo[groupB].size(); j++) { //the leaf nodes that have seqs from group l
 187                                 map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupB][j]].pcount.find(groupB);
 188                                 int numSeqsInGroupL = it->second;
 189
 190                                 double sum = getLengthToRoot(t, t->groupNodeInfo[groupB][j], groupA, groupB);
 191                                 double weightedSum = ((numSeqsInGroupL * sum) / (double)tmap->seqsPerGroup[groupB]);
 192
 193                                 D[count] += weightedSum;
 194                         }
 195                         count++;
 196                 }
 197
 198                 //calculate u for the group comb
 199                 for (int h = start; h < (start+num); h++) {
 200                         //report progress
 201                         //m->mothurOut("Processing combo: " + toString(h)); m->mothurOutEndLine();
 202
 203                         string groupA = namesOfGroupCombos[h][0];
 204                         string groupB = namesOfGroupCombos[h][1];
 205
 206                         //calculate u for the group comb
 207                         for(int i=0;i<t->getNumNodes();i++){
 208
 209                                 if (m->control_pressed) { return data; }
 210
 211                                 double u;
 212                                 //int pcountSize = 0;
 213                                 //does this node have descendants from groupA
 214                                 it = t->tree[i].pcount.find(groupA);
 215                                 //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 216                                 if (it != t->tree[i].pcount.end()) {
 217                                         u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
 218                                 }else { u = 0.00; }
 219
 220
 221                                 //does this node have descendants from group l
 222                                 it = t->tree[i].pcount.find(groupB);
 223
 224                                 //if it does subtract their percentage from u
 225                                 if (it != t->tree[i].pcount.end()) {
 226                                         u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 227                                 }
 228
 229                                 if (includeRoot) {
 230                                         if (t->tree[i].getBranchLength() != -1) {
 231                                                 u = abs(u * t->tree[i].getBranchLength());
 232                                                 WScore[(groupA+groupB)] += u;
 233                                         }
 234                                 }else {
 235                                         //if this is not the root then add it
 236                                         if (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0) {
 237                                                 if (t->tree[i].getBranchLength() != -1) {
 238                                                         u = abs(u * t->tree[i].getBranchLength());
 239                                                         WScore[(groupA+groupB)] += u;
 240                                                 }
 241                                         }
 242                                 }
 243                         }
 244
 245                 }
 246
 247                 /********************************************************/
 248                 //calculate weighted score for the group combination
 249                 double UN;
 250                 count = 0;
 251                 for (int h = start; h < (start+num); h++) {
 252                         UN = (WScore[namesOfGroupCombos[h][0]+namesOfGroupCombos[h][1]] / D[count]);
 253                         if (isnan(UN) || isinf(UN)) { UN = 0; }
 254                         results.push_back(UN);
 255                         count++;
 256                 }
 257
 258                 return results;
 259         }
 260         catch(exception& e) {
 261                 m->errorOut(e, "Weighted", "driver");
 262                 exit(1);
 263         }
 264 }
 265 /**************************************************************************************************/
 266 EstOutput Weighted::getValues(Tree* t, string groupA, string groupB) {
 267  try {
 268
 269                 data.clear(); //clear out old values
 270
 271                 if (m->control_pressed) { return data; }
 272
 273                 //initialize weighted score
 274                 WScore[(groupA+groupB)] = 0.0;
 275                 double D = 0.0;
 276                 set<int> validBranches;
 277
 278                 vector<string> groups; groups.push_back(groupA); groups.push_back(groupB);
 279
 280                 //adding the wieghted sums from group i
 281                 for (int j = 0; j < t->groupNodeInfo[groups[0]].size(); j++) { //the leaf nodes that have seqs from group i
 282                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[0]][j]].pcount.find(groups[0]);
 283                         int numSeqsInGroupI = it->second;
 284
 285                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[0]][j], groups[0], groups[1]);
 286                         double weightedSum = ((numSeqsInGroupI * sum) / (double)tmap->seqsPerGroup[groups[0]]);
 287
 288                         D += weightedSum;
 289                 }
 290
 291                 //adding the wieghted sums from group l
 292                 for (int j = 0; j < t->groupNodeInfo[groups[1]].size(); j++) { //the leaf nodes that have seqs from group l
 293                         map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[1]][j]].pcount.find(groups[1]);
 294                         int numSeqsInGroupL = it->second;
 295
 296                         double sum = getLengthToRoot(t, t->groupNodeInfo[groups[1]][j], groups[0], groups[1]);
 297                         double weightedSum = ((numSeqsInGroupL * sum) / (double)tmap->seqsPerGroup[groups[1]]);
 298
 299                         D += weightedSum;
 300                 }
 301
 302                 //calculate u for the group comb
 303                 for(int i=0;i<t->getNumNodes();i++){
 304
 305                         if (m->control_pressed) { return data; }
 306
 307                         double u;
 308                         //int pcountSize = 0;
 309                         //does this node have descendants from groupA
 310                         it = t->tree[i].pcount.find(groupA);
 311                         //if it does u = # of its descendants with a certain group / total number in tree with a certain group
 312                         if (it != t->tree[i].pcount.end()) {
 313                                 u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
 314                         }else { u = 0.00; }
 315
 316
 317                         //does this node have descendants from group l
 318                         it = t->tree[i].pcount.find(groupB);
 319                         //if it does subtract their percentage from u
 320                         if (it != t->tree[i].pcount.end()) {
 321                                 u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 322                         }
 323
 324                         if (includeRoot) {
 325                                 if (t->tree[i].getBranchLength() != -1) {
 326                                         u = abs(u * t->tree[i].getBranchLength());
 327                                         WScore[(groupA+groupB)] += u;
 328                                 }
 329                         }else{
 330                                 //if this is not the root then add it
 331                                 if (rootForGrouping[groups].count(i) == 0) {
 332                                         if (t->tree[i].getBranchLength() != -1) {
 333                                                 u = abs(u * t->tree[i].getBranchLength());
 334                                                 WScore[(groupA+groupB)] += u;
 335                                         }
 336                                 }
 337                         }
 338                 }
 339                 /********************************************************/
 340
 341                 //calculate weighted score for the group combination
 342                 double UN;
 343                 UN = (WScore[(groupA+groupB)] / D);
 344
 345                 if (isnan(UN) || isinf(UN)) { UN = 0; }
 346                 data.push_back(UN);
 347
 348                 return data;
 349         }
 350         catch(exception& e) {
 351                 m->errorOut(e, "Weighted", "getValues");
 352                 exit(1);
 353         }
 354 }
 355 /**************************************************************************************************/
 356 double Weighted::getLengthToRoot(Tree* t, int v, string groupA, string groupB) {
 357         try {
 358
 359                 double sum = 0.0;
 360                 int index = v;
 361
 362                 //you are a leaf
 363                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 364                 double tempTotal = 0.0;
 365                 index = t->tree[index].getParent();
 366
 367                 vector<string> grouping; grouping.push_back(groupA); grouping.push_back(groupB);
 368
 369                 rootForGrouping[grouping].insert(index);
 370
 371                 //while you aren't at root
 372                 while(t->tree[index].getParent() != -1){
 373
 374                         if (m->control_pressed) {  return sum; }
 375
 376                         int parent = t->tree[index].getParent();
 377
 378                         if (includeRoot) { //add everyone
 379                                 if(t->tree[index].getBranchLength() != -1){     sum += abs(t->tree[index].getBranchLength());   }
 380                         }else {
 381
 382                                 //am I the root for this grouping? if so I want to stop "early"
 383                                 //does my sibling have descendants from the users groups?
 384                                 int lc = t->tree[parent].getLChild();
 385                                 int rc = t->tree[parent].getRChild();
 386
 387                                 int sib = lc;
 388                                 if (lc == index) { sib = rc; }
 389
 390                                 map<string, int>::iterator itGroup;
 391                                 int pcountSize = 0;
 392                                 itGroup = t->tree[sib].pcount.find(groupA);
 393                                 if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 394                                 itGroup = t->tree[sib].pcount.find(groupB);
 395                                 if (itGroup != t->tree[sib].pcount.end()) { pcountSize++;  }
 396
 397                                 //if yes, I am not the root so add me
 398                                 if (pcountSize != 0) {
 399                                         if (t->tree[index].getBranchLength() != -1) {
 400                                                 sum += abs(t->tree[index].getBranchLength()) + tempTotal;
 401                                                 tempTotal = 0.0;
 402                                         }else {
 403                                                 sum += tempTotal;
 404                                                 tempTotal = 0.0;
 405                                         }
 406                                         rootForGrouping[grouping].clear();
 407                                         rootForGrouping[grouping].insert(parent);
 408                                 }else { //if no, I may be the root so add my br to tempTotal until I am proven innocent
 409                                         if (t->tree[index].getBranchLength() != -1) {
 410                                                 tempTotal += abs(t->tree[index].getBranchLength());
 411                                         }
 412                                 }
 413                         }
 414
 415                         index = parent;
 416                 }
 417
 418                 //get all nodes above the root to add so we don't add their u values above
 419                 index = *(rootForGrouping[grouping].begin());
 420
 421                 while(t->tree[index].getParent() != -1){
 422                         int parent = t->tree[index].getParent();
 423                         rootForGrouping[grouping].insert(parent);
 424                         index = parent;
 425                 }
 426                 return sum;
 427         }
 428         catch(exception& e) {
 429                 m->errorOut(e, "Weighted", "getBranchLengthSums");
 430                 exit(1);
 431         }
 432 }
 433 /**************************************************************************************************/
 434
 435