unweighted.cpp

   1 /*
   2  *  unweighted.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 "unweighted.h"
  11
  12 /**************************************************************************************************/
  13
  14 EstOutput Unweighted::getValues(Tree* t, int p, string o) {
  15         try {
  16                 globaldata = GlobalData::getInstance();
  17                 processors = p;
  18                 outputDir = o;
  19
  20                 //if the users enters no groups then give them the score of all groups
  21                 int numGroups = globaldata->Groups.size();
  22
  23                 //calculate number of comparsions
  24                 int numComp = 0;
  25                 vector< vector<string> > namesOfGroupCombos;
  26                 for (int r=0; r<numGroups; r++) {
  27                         for (int l = 0; l < r; l++) {
  28                                 numComp++;
  29                                 vector<string> groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]);
  30                                 namesOfGroupCombos.push_back(groups);
  31                         }
  32                 }
  33
  34                 if (numComp != 1) {
  35                         vector<string> groups;
  36                         if (numGroups == 0) {
  37                                 //get score for all users groups
  38                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
  39                                         if (tmap->namesOfGroups[i] != "xxx") {
  40                                                 groups.push_back(tmap->namesOfGroups[i]);
  41                                         }
  42                                 }
  43                                 namesOfGroupCombos.push_back(groups);
  44                         }else {
  45                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
  46                                         groups.push_back(globaldata->Groups[i]);
  47                                 }
  48                                 namesOfGroupCombos.push_back(groups);
  49                         }
  50                 }
  51
  52                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  53                         if(processors == 1){
  54                                 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
  55                         }else{
  56                                 int numPairs = namesOfGroupCombos.size();
  57
  58                                 int numPairsPerProcessor = numPairs / processors;
  59
  60                                 for (int i = 0; i < processors; i++) {
  61                                         int startPos = i * numPairsPerProcessor;
  62
  63                                         if(i == processors - 1){
  64                                                 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
  65                                         }
  66
  67                                         lines.push_back(linePair(startPos, numPairsPerProcessor));
  68                                 }
  69                                 data = createProcesses(t, namesOfGroupCombos);
  70                                 lines.clear();
  71                         }
  72                 #else
  73                         data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
  74                 #endif
  75
  76                 return data;
  77         }
  78         catch(exception& e) {
  79                 m->errorOut(e, "Unweighted", "getValues");
  80                 exit(1);
  81         }
  82 }
  83 /**************************************************************************************************/
  84
  85 EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos) {
  86         try {
  87 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  88                 int process = 1;
  89                 vector<int> processIDS;
  90
  91                 EstOutput results;
  92
  93                 //loop through and create all the processes you want
  94                 while (process != processors) {
  95                         int pid = fork();
  96
  97                         if (pid > 0) {
  98                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
  99                                 process++;
 100                         }else if (pid == 0){
 101                                 EstOutput myresults;
 102                                 myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num);
 103
 104                                 if (m->control_pressed) { exit(0); }
 105
 106                                 m->mothurOut("Merging results."); m->mothurOutEndLine();
 107
 108                                 //pass numSeqs to parent
 109                                 ofstream out;
 110                                 string tempFile = outputDir + toString(getpid()) + ".unweighted.results.temp";
 111                                 m->openOutputFile(tempFile, out);
 112                                 out << myresults.size() << endl;
 113                                 for (int i = 0; i < myresults.size(); i++) {  out << myresults[i] << '\t';  } out << endl;
 114                                 out.close();
 115
 116                                 exit(0);
 117                         }else {
 118                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 119                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 120                                 exit(0);
 121                         }
 122                 }
 123
 124                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num);
 125
 126                 //force parent to wait until all the processes are done
 127                 for (int i=0;i<(processors-1);i++) {
 128                         int temp = processIDS[i];
 129                         wait(&temp);
 130                 }
 131
 132                 if (m->control_pressed) { return results; }
 133
 134                 //get data created by processes
 135                 for (int i=0;i<(processors-1);i++) {
 136                         ifstream in;
 137                         string s = outputDir + toString(processIDS[i]) + ".unweighted.results.temp";
 138                         m->openInputFile(s, in);
 139
 140                         //get quantiles
 141                         if (!in.eof()) {
 142                                 int num;
 143                                 in >> num; m->gobble(in);
 144
 145                                 if (m->control_pressed) { break; }
 146
 147                                 double w;
 148                                 for (int j = 0; j < num; j++) {
 149                                         in >> w;
 150                                         results.push_back(w);
 151                                 }
 152                                 m->gobble(in);
 153                         }
 154                         in.close();
 155                         remove(s.c_str());
 156                 }
 157
 158                 m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
 159
 160                 return results;
 161 #endif
 162         }
 163         catch(exception& e) {
 164                 m->errorOut(e, "Unweighted", "createProcesses");
 165                 exit(1);
 166         }
 167 }
 168 /**************************************************************************************************/
 169 EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num) {
 170  try {
 171
 172                 EstOutput results; results.resize(num);
 173
 174                 int count = 0;
 175                 int total = num;
 176                 int twentyPercent = (total * 0.20);
 177                 if (twentyPercent == 0) { twentyPercent = 1; }
 178
 179
 180                 for (int h = start; h < (start+num); h++) {
 181         //cout << namesOfGroupCombos[h][0] << '\t' << namesOfGroupCombos[h][1] << endl;
 182                         if (m->control_pressed) { return results; }
 183
 184                         double UniqueBL=0.0000;  //a branch length is unique if it's chidren are from the same group
 185                         double totalBL = 0.00;  //all branch lengths
 186                         double UW = 0.00;               //Unweighted Value = UniqueBL / totalBL;
 187
 188                         //find a node that belongs to one of the groups in this combo
 189                         int nodeBelonging = -1;
 190                         for (int g = 0; g < namesOfGroupCombos[h].size(); g++) {
 191                                 if (t->groupNodeInfo[namesOfGroupCombos[h][g]].size() != 0) { nodeBelonging = t->groupNodeInfo[namesOfGroupCombos[h][g]][0]; break; }
 192                         }
 193
 194                         //sanity check
 195                         if (nodeBelonging == -1) {
 196                                 m->mothurOut("[WARNING]: cannot find a nodes in the tree from grouping ");
 197                                 for (int g = 0; g < namesOfGroupCombos[h].size()-1; g++) { m->mothurOut(namesOfGroupCombos[h][g] + "-"); }
 198                                 m->mothurOut(namesOfGroupCombos[h][namesOfGroupCombos[h].size()-1]);
 199                                 m->mothurOut(", skipping."); m->mothurOutEndLine(); results[count] = UW;
 200                         }else{
 201
 202                                 getRoot(t, nodeBelonging, namesOfGroupCombos[h]);
 203
 204                                 for(int i=0;i<t->getNumNodes();i++){
 205
 206                                         if (m->control_pressed) {  return data; }
 207
 208                                         //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
 209                                         //pcountSize = 2, not unique to one group
 210                                         //pcountSize = 1, unique to one group
 211
 212                                         int pcountSize = 0;
 213                                         for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
 214                                                 map<string, int>::iterator itGroup = t->tree[i].pcount.find(namesOfGroupCombos[h][j]);
 215                                                 if (itGroup != t->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } }
 216                                         }
 217                                         //
 218                                         //unique calc
 219                                         if (pcountSize == 0) { }
 220                                         else if ((t->tree[i].getBranchLength() != -1) && (pcountSize == 1) && (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0)) { //you have a unique branch length and you are not the root
 221                                                 UniqueBL += abs(t->tree[i].getBranchLength());
 222                                         }
 223
 224                                         //total calc
 225                                         if (pcountSize == 0) { }
 226                                         else if ((t->tree[i].getBranchLength() != -1) && (pcountSize != 0) && (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0)) { //you have a branch length and you are not the root
 227                                                 totalBL += abs(t->tree[i].getBranchLength());
 228                                         }
 229
 230                                 }
 231         cout << UniqueBL << '\t' << totalBL << endl;
 232                                 UW = (UniqueBL / totalBL);
 233
 234                                 if (isnan(UW) || isinf(UW)) { UW = 0; }
 235
 236                                 results[count] = UW;
 237                         }
 238                         count++;
 239
 240                         //report progress
 241                         if((count % twentyPercent) == 0) {      float tempOut = (count / (float)total); if (isnan(tempOut) || isinf(tempOut)) { tempOut = 0.0; } m->mothurOut("Percentage complete: " + toString((int(tempOut) * 100.0))); m->mothurOutEndLine();       }
 242                 }
 243
 244                 //report progress
 245                 if((count % twentyPercent) != 0) {      float tempOut = (count / (float)total); if (isnan(tempOut) || isinf(tempOut)) { tempOut = 0.0; } m->mothurOut("Percentage complete: " + toString((int(tempOut) * 100.0))); m->mothurOutEndLine();       }
 246
 247                 return results;
 248         }
 249         catch(exception& e) {
 250                 m->errorOut(e, "Unweighted", "driver");
 251                 exit(1);
 252         }
 253 }
 254 /**************************************************************************************************/
 255
 256 EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, string o) {
 257  try {
 258                 globaldata = GlobalData::getInstance();
 259                 processors = p;
 260                 outputDir = o;
 261
 262
 263                 //if the users enters no groups then give them the score of all groups
 264                 int numGroups = globaldata->Groups.size();
 265
 266                 //calculate number of comparsions
 267                 int numComp = 0;
 268                 vector< vector<string> > namesOfGroupCombos;
 269                 for (int r=0; r<numGroups; r++) {
 270                         for (int l = 0; l < r; l++) {
 271                                 numComp++;
 272                                 vector<string> groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]);
 273                                 namesOfGroupCombos.push_back(groups);
 274                         }
 275                 }
 276
 277                 if (numComp != 1) {
 278                         vector<string> groups;
 279                         if (numGroups == 0) {
 280                                 //get score for all users groups
 281                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 282                                         if (tmap->namesOfGroups[i] != "xxx") {
 283                                                 groups.push_back(tmap->namesOfGroups[i]);
 284                                         }
 285                                 }
 286                                 namesOfGroupCombos.push_back(groups);
 287                         }else {
 288                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 289                                         groups.push_back(globaldata->Groups[i]);
 290                                 }
 291                                 namesOfGroupCombos.push_back(groups);
 292                         }
 293                 }
 294
 295                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 296                         if(processors == 1){
 297                                 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
 298                         }else{
 299                                 int numPairs = namesOfGroupCombos.size();
 300
 301                                 int numPairsPerProcessor = numPairs / processors;
 302
 303                                 for (int i = 0; i < processors; i++) {
 304                                         int startPos = i * numPairsPerProcessor;
 305                                         if(i == processors - 1){
 306                                                 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
 307                                         }
 308                                         lines.push_back(linePair(startPos, numPairsPerProcessor));
 309                                 }
 310
 311                                 data = createProcesses(t, namesOfGroupCombos, true);
 312
 313                                 lines.clear();
 314                         }
 315                 #else
 316                         data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
 317                 #endif
 318
 319                 return data;
 320         }
 321         catch(exception& e) {
 322                 m->errorOut(e, "Unweighted", "getValues");
 323                 exit(1);
 324         }
 325 }
 326 /**************************************************************************************************/
 327
 328 EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, bool usingGroups) {
 329         try {
 330 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 331                 int process = 1;
 332                 vector<int> processIDS;
 333
 334                 EstOutput results;
 335
 336                 //loop through and create all the processes you want
 337                 while (process != processors) {
 338                         int pid = fork();
 339
 340                         if (pid > 0) {
 341                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 342                                 process++;
 343                         }else if (pid == 0){
 344                                 EstOutput myresults;
 345                                 myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, usingGroups);
 346
 347                                 if (m->control_pressed) { exit(0); }
 348
 349                                 //pass numSeqs to parent
 350                                 ofstream out;
 351                                 string tempFile = outputDir + toString(getpid()) + ".unweighted.results.temp";
 352                                 m->openOutputFile(tempFile, out);
 353                                 out << myresults.size() << endl;
 354                                 for (int i = 0; i < myresults.size(); i++) {  out << myresults[i] << '\t';  } out << endl;
 355                                 out.close();
 356
 357                                 exit(0);
 358                         }else {
 359                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 360                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 361                                 exit(0);
 362                         }
 363                 }
 364
 365                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, usingGroups);
 366
 367                 //force parent to wait until all the processes are done
 368                 for (int i=0;i<(processors-1);i++) {
 369                         int temp = processIDS[i];
 370                         wait(&temp);
 371                 }
 372
 373                 if (m->control_pressed) { return results; }
 374
 375                 //get data created by processes
 376                 for (int i=0;i<(processors-1);i++) {
 377                         ifstream in;
 378                         string s = outputDir + toString(processIDS[i]) + ".unweighted.results.temp";
 379                         m->openInputFile(s, in);
 380
 381                         //get quantiles
 382                         if (!in.eof()) {
 383                                 int num;
 384                                 in >> num; m->gobble(in);
 385
 386                                 if (m->control_pressed) { break; }
 387
 388                                 double w;
 389                                 for (int j = 0; j < num; j++) {
 390                                         in >> w;
 391
 392                                         results.push_back(w);
 393                                 }
 394                                 m->gobble(in);
 395                         }
 396                         in.close();
 397                         remove(s.c_str());
 398                 }
 399
 400                 return results;
 401 #endif
 402         }
 403         catch(exception& e) {
 404                 m->errorOut(e, "Unweighted", "createProcesses");
 405                 exit(1);
 406         }
 407 }
 408 /**************************************************************************************************/
 409 EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, bool usingGroups) {
 410  try {
 411
 412                 EstOutput results; results.resize(num);
 413
 414                 int count = 0;
 415
 416                 Tree* copyTree = new Tree;
 417
 418                 for (int h = start; h < (start+num); h++) {
 419
 420                         if (m->control_pressed) { return results; }
 421
 422                         //copy random tree passed in
 423                         copyTree->getCopy(t);
 424
 425                         //swap labels in the groups you want to compare
 426                         copyTree->assembleRandomUnifracTree(namesOfGroupCombos[h]);
 427
 428                         double UniqueBL=0.0000;  //a branch length is unique if it's chidren are from the same group
 429                         double totalBL = 0.00;  //all branch lengths
 430                         double UW = 0.00;               //Unweighted Value = UniqueBL / totalBL;
 431                         //find a node that belongs to one of the groups in this combo
 432                         int nodeBelonging = -1;
 433                         for (int g = 0; g < namesOfGroupCombos[h].size(); g++) {
 434                                 if (copyTree->groupNodeInfo[namesOfGroupCombos[h][g]].size() != 0) { nodeBelonging = copyTree->groupNodeInfo[namesOfGroupCombos[h][g]][0]; break; }
 435                         }
 436
 437                         //sanity check
 438                         if (nodeBelonging == -1) {
 439                                 m->mothurOut("[WARNING]: cannot find a nodes in the tree from grouping ");
 440                                 for (int g = 0; g < namesOfGroupCombos[h].size()-1; g++) { m->mothurOut(namesOfGroupCombos[h][g] + "-"); }
 441                                 m->mothurOut(namesOfGroupCombos[h][namesOfGroupCombos[h].size()-1]);
 442                                 m->mothurOut(", skipping."); m->mothurOutEndLine(); results[count] = UW;
 443                         }else{
 444
 445                                 getRoot(copyTree, nodeBelonging, namesOfGroupCombos[h]);
 446
 447                                 for(int i=0;i<copyTree->getNumNodes();i++){
 448
 449                                         if (m->control_pressed) {  return data; }
 450
 451                                         //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
 452                                         //pcountSize = 2, not unique to one group
 453                                         //pcountSize = 1, unique to one group
 454
 455                                         int pcountSize = 0;
 456                                         for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
 457                                                 map<string, int>::iterator itGroup = copyTree->tree[i].pcount.find(namesOfGroupCombos[h][j]);
 458                                                 if (itGroup != copyTree->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } }
 459                                         }
 460
 461                                         //unique calc
 462                                         if (pcountSize == 0) { }
 463                                         else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize == 1) && (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0)) { //you have a unique branch length and you are not the root
 464                                                 UniqueBL += abs(copyTree->tree[i].getBranchLength());
 465                                         }
 466
 467                                         //total calc
 468                                         if (pcountSize == 0) { }
 469                                         else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize != 0) && (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0)) { //you have a branch length and you are not the root
 470                                                 totalBL += abs(copyTree->tree[i].getBranchLength());
 471                                         }
 472
 473                                 }
 474                                 //cout << UniqueBL << '\t' << totalBL << endl;
 475                                 UW = (UniqueBL / totalBL);
 476
 477                                 if (isnan(UW) || isinf(UW)) { UW = 0; }
 478
 479                                 results[count] = UW;
 480                         }
 481                         count++;
 482
 483                 }
 484
 485                 delete copyTree;
 486
 487                 return results;
 488         }
 489         catch(exception& e) {
 490                 m->errorOut(e, "Unweighted", "driver");
 491                 exit(1);
 492         }
 493 }
 494 /**************************************************************************************************/
 495 int Unweighted::getRoot(Tree* t, int v, vector<string> grouping) {
 496         try {
 497                 //you are a leaf so get your parent
 498                 int index = t->tree[index].getParent();
 499
 500                 //my parent is a potential root
 501                 rootForGrouping[grouping].insert(index);
 502
 503                 //while you aren't at root
 504                 while(t->tree[index].getParent() != -1){
 505
 506                         if (m->control_pressed) {  return 0; }
 507
 508                         //am I the root for this grouping? if so I want to stop "early"
 509                         //does my sibling have descendants from the users groups?
 510                         //if so I am not the root
 511                         int parent = t->tree[index].getParent();
 512                         int lc = t->tree[parent].getLChild();
 513                         int rc = t->tree[parent].getRChild();
 514
 515                         int sib = lc;
 516                         if (lc == index) { sib = rc; }
 517
 518                         map<string, int>::iterator itGroup;
 519                         int pcountSize = 0;
 520                         for (int j = 0; j < grouping.size(); j++) {
 521                                 map<string, int>::iterator itGroup = t->tree[sib].pcount.find(grouping[j]);
 522                                 if (itGroup != t->tree[sib].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } }
 523                         }
 524
 525                         //if yes, I am not the root
 526                         if (pcountSize != 0) {
 527                                 rootForGrouping[grouping].clear();
 528                                 rootForGrouping[grouping].insert(parent);
 529                         }
 530
 531                         index = parent;
 532                 }
 533
 534                 //get all nodes above the root to add so we don't add their u values above
 535                 index = *(rootForGrouping[grouping].begin());
 536                 while(t->tree[index].getParent() != -1){
 537                         int parent = t->tree[index].getParent();
 538                         rootForGrouping[grouping].insert(parent);
 539                         cout << parent << " in root" << endl;
 540                         index = parent;
 541                 }
 542
 543                 return 0;
 544         }
 545         catch(exception& e) {
 546                 m->errorOut(e, "Unweighted", "getRoot");
 547                 exit(1);
 548         }
 549 }
 550 /**************************************************************************************************/
 551
 552