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
  70                                 data = createProcesses(t, namesOfGroupCombos);
  71                                 lines.clear();
  72                         }
  73                 #else
  74                         data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
  75                 #endif
  76
  77                 return data;
  78         }
  79         catch(exception& e) {
  80                 m->errorOut(e, "Unweighted", "getValues");
  81                 exit(1);
  82         }
  83 }
  84 /**************************************************************************************************/
  85
  86 EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos) {
  87         try {
  88 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  89                 int process = 1;
  90                 int num = 0;
  91                 vector<int> processIDS;
  92
  93                 EstOutput results;
  94
  95                 //loop through and create all the processes you want
  96                 while (process != processors) {
  97                         int pid = fork();
  98
  99                         if (pid > 0) {
 100                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 101                                 process++;
 102                         }else if (pid == 0){
 103                                 EstOutput myresults;
 104                                 myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num);
 105
 106                                 if (m->control_pressed) { exit(0); }
 107
 108                                 m->mothurOut("Merging results."); m->mothurOutEndLine();
 109
 110                                 //pass numSeqs to parent
 111                                 ofstream out;
 112                                 string tempFile = outputDir + toString(getpid()) + ".unweighted.results.temp";
 113                                 m->openOutputFile(tempFile, out);
 114                                 out << myresults.size() << endl;
 115                                 for (int i = 0; i < myresults.size(); i++) {  out << myresults[i] << '\t';  } out << endl;
 116                                 out.close();
 117
 118                                 exit(0);
 119                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 120                 }
 121
 122                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num);
 123
 124                 //force parent to wait until all the processes are done
 125                 for (int i=0;i<(processors-1);i++) {
 126                         int temp = processIDS[i];
 127                         wait(&temp);
 128                 }
 129
 130                 if (m->control_pressed) { return results; }
 131
 132                 //get data created by processes
 133                 for (int i=0;i<(processors-1);i++) {
 134                         ifstream in;
 135                         string s = outputDir + toString(processIDS[i]) + ".unweighted.results.temp";
 136                         m->openInputFile(s, in);
 137
 138                         //get quantiles
 139                         if (!in.eof()) {
 140                                 int num;
 141                                 in >> num; m->gobble(in);
 142
 143                                 if (m->control_pressed) { break; }
 144
 145                                 double w;
 146                                 for (int j = 0; j < num; j++) {
 147                                         in >> w;
 148                                         results.push_back(w);
 149                                 }
 150                                 m->gobble(in);
 151                         }
 152                         in.close();
 153                         remove(s.c_str());
 154                 }
 155
 156                 m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
 157
 158                 return results;
 159 #endif
 160         }
 161         catch(exception& e) {
 162                 m->errorOut(e, "Unweighted", "createProcesses");
 163                 exit(1);
 164         }
 165 }
 166 /**************************************************************************************************/
 167 EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num) {
 168  try {
 169
 170                 EstOutput results; results.resize(num);
 171
 172                 int count = 0;
 173                 int numLeaves = t->getNumLeaves();
 174                 int total = num;
 175                 int twentyPercent = (total * 0.20);
 176                 if (twentyPercent == 0) { twentyPercent = 1; }
 177
 178
 179                 for (int h = start; h < (start+num); h++) {
 180
 181                         if (m->control_pressed) { return results; }
 182
 183                         double UniqueBL=0.0000;  //a branch length is unique if it's chidren are from the same group
 184                         double totalBL = 0.00;  //all branch lengths
 185                         double UW = 0.00;               //Unweighted Value = UniqueBL / totalBL;
 186                         map<int, double> tempTotals; //maps node to total Branch Length
 187                         map<int, int> nodePcountSize; //maps node to pcountSize
 188                         map<int, int>::iterator itCount;
 189
 190                         for(int i=0;i<t->getNumNodes();i++){
 191
 192                                 if (m->control_pressed) {  return data; }
 193
 194                                 //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
 195                                 //pcountSize = 2, not unique to one group
 196                                 //pcountSize = 1, unique to one group
 197
 198                                 int pcountSize = 0;
 199                                 for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
 200                                         map<string, int>::iterator itGroup = t->tree[i].pcount.find(namesOfGroupCombos[h][j]);
 201                                         if (itGroup != t->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } }
 202                                 }
 203
 204                                 nodePcountSize[i] = pcountSize;
 205
 206                                 if (pcountSize == 0) { ; }
 207                                 else if ((t->tree[i].getBranchLength() != -1) && (pcountSize == 1)) {  UniqueBL += abs(t->tree[i].getBranchLength()); }
 208
 209                                 //if you are a leaf from a users group add to total
 210                                 if (i < numLeaves) {
 211                                         if ((t->tree[i].getBranchLength() != -1) && pcountSize != 0) {
 212                                                 totalBL += abs(t->tree[i].getBranchLength());
 213                                         }
 214                                         tempTotals[i] = 0.0;  //we don't care about you, or we have already added you
 215                                 }else{ //if you are not a leaf
 216                                         //do both your chidren have have descendants from the users groups?
 217                                         int lc = t->tree[i].getLChild();
 218                                         int rc = t->tree[i].getRChild();
 219
 220                                         //if yes, add your childrens tempTotals
 221                                         if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
 222                                                 totalBL += tempTotals[lc] + tempTotals[rc];
 223                                                 if (t->tree[i].getBranchLength() != -1) {
 224                                                         tempTotals[i] = abs(t->tree[i].getBranchLength());
 225                                                 }else {
 226                                                         tempTotals[i] = 0.0;
 227                                                 }
 228                                         }else { //if no, your tempTotal is your childrens temp totals + your branch length
 229                                                 tempTotals[i] = tempTotals[lc] + tempTotals[rc] + abs(t->tree[i].getBranchLength());
 230                                         }
 231                                 }
 232                         }
 233         cout << UniqueBL << '\t' << totalBL << endl;
 234                         UW = (UniqueBL / totalBL);
 235
 236                         if (isnan(UW) || isinf(UW)) { UW = 0; }
 237
 238                         results[count] = UW;
 239                         count++;
 240
 241                         //report progress
 242                         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();       }
 243                 }
 244
 245                 //report progress
 246                 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();       }
 247
 248                 return results;
 249         }
 250         catch(exception& e) {
 251                 m->errorOut(e, "Unweighted", "driver");
 252                 exit(1);
 253         }
 254 }
 255 /**************************************************************************************************/
 256
 257 EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, string o) {
 258  try {
 259                 globaldata = GlobalData::getInstance();
 260                 processors = p;
 261                 outputDir = o;
 262
 263
 264                 //if the users enters no groups then give them the score of all groups
 265                 int numGroups = globaldata->Groups.size();
 266
 267                 //calculate number of comparsions
 268                 int numComp = 0;
 269                 vector< vector<string> > namesOfGroupCombos;
 270                 for (int r=0; r<numGroups; r++) {
 271                         for (int l = 0; l < r; l++) {
 272                                 numComp++;
 273                                 vector<string> groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]);
 274                                 namesOfGroupCombos.push_back(groups);
 275                         }
 276                 }
 277
 278                 if (numComp != 1) {
 279                         vector<string> groups;
 280                         if (numGroups == 0) {
 281                                 //get score for all users groups
 282                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 283                                         if (tmap->namesOfGroups[i] != "xxx") {
 284                                                 groups.push_back(tmap->namesOfGroups[i]);
 285                                         }
 286                                 }
 287                                 namesOfGroupCombos.push_back(groups);
 288                         }else {
 289                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 290                                         groups.push_back(globaldata->Groups[i]);
 291                                 }
 292                                 namesOfGroupCombos.push_back(groups);
 293                         }
 294                 }
 295
 296                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 297                         if(processors == 1){
 298                                 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
 299                         }else{
 300                                 int numPairs = namesOfGroupCombos.size();
 301
 302                                 int numPairsPerProcessor = numPairs / processors;
 303
 304                                 for (int i = 0; i < processors; i++) {
 305                                         int startPos = i * numPairsPerProcessor;
 306                                         if(i == processors - 1){
 307                                                 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
 308                                         }
 309                                         lines.push_back(linePair(startPos, numPairsPerProcessor));
 310                                 }
 311
 312                                 data = createProcesses(t, namesOfGroupCombos, true);
 313
 314                                 lines.clear();
 315                         }
 316                 #else
 317                         data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
 318                 #endif
 319
 320                 return data;
 321         }
 322         catch(exception& e) {
 323                 m->errorOut(e, "Unweighted", "getValues");
 324                 exit(1);
 325         }
 326 }
 327 /**************************************************************************************************/
 328
 329 EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, bool usingGroups) {
 330         try {
 331 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 332                 int process = 1;
 333                 int num = 0;
 334                 vector<int> processIDS;
 335
 336                 EstOutput results;
 337
 338                 //loop through and create all the processes you want
 339                 while (process != processors) {
 340                         int pid = fork();
 341
 342                         if (pid > 0) {
 343                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 344                                 process++;
 345                         }else if (pid == 0){
 346                                 EstOutput myresults;
 347                                 myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, usingGroups);
 348
 349                                 if (m->control_pressed) { exit(0); }
 350
 351                                 //pass numSeqs to parent
 352                                 ofstream out;
 353                                 string tempFile = outputDir + toString(getpid()) + ".unweighted.results.temp";
 354                                 m->openOutputFile(tempFile, out);
 355                                 out << myresults.size() << endl;
 356                                 for (int i = 0; i < myresults.size(); i++) {  out << myresults[i] << '\t';  } out << endl;
 357                                 out.close();
 358
 359                                 exit(0);
 360                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 361                 }
 362
 363                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, usingGroups);
 364
 365                 //force parent to wait until all the processes are done
 366                 for (int i=0;i<(processors-1);i++) {
 367                         int temp = processIDS[i];
 368                         wait(&temp);
 369                 }
 370
 371                 if (m->control_pressed) { return results; }
 372
 373                 //get data created by processes
 374                 for (int i=0;i<(processors-1);i++) {
 375                         ifstream in;
 376                         string s = outputDir + toString(processIDS[i]) + ".unweighted.results.temp";
 377                         m->openInputFile(s, in);
 378
 379                         //get quantiles
 380                         if (!in.eof()) {
 381                                 int num;
 382                                 in >> num; m->gobble(in);
 383
 384                                 if (m->control_pressed) { break; }
 385
 386                                 double w;
 387                                 for (int j = 0; j < num; j++) {
 388                                         in >> w;
 389                                         results.push_back(w);
 390                                 }
 391                                 m->gobble(in);
 392                         }
 393                         in.close();
 394                         remove(s.c_str());
 395                 }
 396
 397                 return results;
 398 #endif
 399         }
 400         catch(exception& e) {
 401                 m->errorOut(e, "Unweighted", "createProcesses");
 402                 exit(1);
 403         }
 404 }
 405 /**************************************************************************************************/
 406 EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, bool usingGroups) {
 407  try {
 408
 409                 EstOutput results; results.resize(num);
 410
 411                 int count = 0;
 412                 int total = num;
 413                 int twentyPercent = (total * 0.20);
 414                 int numLeaves = t->getNumLeaves();
 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                         map<int, double> tempTotals; //maps node to total Branch Length
 432                         map<int, int> nodePcountSize; //maps node to pcountSize
 433
 434                         for(int i=0;i<copyTree->getNumNodes();i++){
 435
 436                                 if (m->control_pressed) {  return data; }
 437
 438                                 //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
 439                                 //pcountSize = 2, not unique to one group
 440                                 //pcountSize = 1, unique to one group
 441
 442                                 int pcountSize = 0;
 443                                 for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
 444                                         map<string, int>::iterator itGroup = copyTree->tree[i].pcount.find(namesOfGroupCombos[h][j]);
 445                                         if (itGroup != copyTree->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } }
 446                                 }
 447
 448                                 nodePcountSize[i] = pcountSize;
 449
 450                                 if (pcountSize == 0) { }
 451                                 else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize == 1)) {  UniqueBL += abs(copyTree->tree[i].getBranchLength());     }
 452
 453                                 //if you are a leaf from a users group add to total
 454                                 if (i < numLeaves) {
 455                                         if ((copyTree->tree[i].getBranchLength() != -1) && pcountSize != 0) {
 456                                                 totalBL += abs(copyTree->tree[i].getBranchLength());
 457                                         }
 458                                         tempTotals[i] = 0.0;  //we don't care about you, or we have already added you
 459                                 }else{ //if you are not a leaf
 460                                         //do both your chidren have have descendants from the users groups?
 461                                         int lc = copyTree->tree[i].getLChild();
 462                                         int rc = copyTree->tree[i].getRChild();
 463
 464                                         //if yes, add your childrens tempTotals
 465                                         if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
 466                                                 totalBL += tempTotals[lc] + tempTotals[rc];
 467                                                 if (copyTree->tree[i].getBranchLength() != -1) {
 468                                                         tempTotals[i] = abs(copyTree->tree[i].getBranchLength());
 469                                                 }else {
 470                                                         tempTotals[i] = 0.0;
 471                                                 }
 472                                         }else { //if no, your tempTotal is your childrens temp totals + your branch length
 473                                                 tempTotals[i] = tempTotals[lc] + tempTotals[rc] + abs(copyTree->tree[i].getBranchLength());
 474                                         }
 475                                 }
 476
 477                         }
 478
 479                         UW = (UniqueBL / totalBL);
 480
 481                         if (isnan(UW) || isinf(UW)) { UW = 0; }
 482
 483                         results[count] = UW;
 484                         count++;
 485
 486                 }
 487
 488                 delete copyTree;
 489
 490                 return results;
 491         }
 492         catch(exception& e) {
 493                 m->errorOut(e, "Unweighted", "driver");
 494                 exit(1);
 495         }
 496 }
 497 /**************************************************************************************************/
 498
 499