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