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 to total your childrens branchLengths and 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                                                 //tempTotals[i] = tempTotals[lc] + tempTotals[rc];
 229                                         }else { //if no, your tempTotal is your childrens temp totals
 230                                                 tempTotals[i] = tempTotals[lc] + tempTotals[rc] + abs(t->tree[i].getBranchLength());
 231                                         }
 232                                 }
 233                         }
 234
 235                         UW = (UniqueBL / totalBL);
 236
 237                         if (isnan(UW) || isinf(UW)) { UW = 0; }
 238
 239                         results[count] = UW;
 240                         count++;
 241
 242                         //report progress
 243                         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();       }
 244                 }
 245
 246                 //report progress
 247                 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();       }
 248
 249                 return results;
 250         }
 251         catch(exception& e) {
 252                 m->errorOut(e, "Unweighted", "driver");
 253                 exit(1);
 254         }
 255 }
 256 /**************************************************************************************************/
 257
 258 EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, string o) {
 259  try {
 260                 globaldata = GlobalData::getInstance();
 261                 processors = p;
 262                 outputDir = o;
 263
 264
 265                 //if the users enters no groups then give them the score of all groups
 266                 int numGroups = globaldata->Groups.size();
 267
 268                 //calculate number of comparsions
 269                 int numComp = 0;
 270                 vector< vector<string> > namesOfGroupCombos;
 271                 for (int r=0; r<numGroups; r++) {
 272                         for (int l = 0; l < r; l++) {
 273                                 numComp++;
 274                                 vector<string> groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]);
 275                                 namesOfGroupCombos.push_back(groups);
 276                         }
 277                 }
 278
 279                 if (numComp != 1) {
 280                         vector<string> groups;
 281                         if (numGroups == 0) {
 282                                 //get score for all users groups
 283                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 284                                         if (tmap->namesOfGroups[i] != "xxx") {
 285                                                 groups.push_back(tmap->namesOfGroups[i]);
 286                                         }
 287                                 }
 288                                 namesOfGroupCombos.push_back(groups);
 289                         }else {
 290                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 291                                         groups.push_back(globaldata->Groups[i]);
 292                                 }
 293                                 namesOfGroupCombos.push_back(groups);
 294                         }
 295                 }
 296
 297                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 298                         if(processors == 1){
 299                                 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
 300                         }else{
 301                                 int numPairs = namesOfGroupCombos.size();
 302
 303                                 int numPairsPerProcessor = numPairs / processors;
 304
 305                                 for (int i = 0; i < processors; i++) {
 306                                         int startPos = i * numPairsPerProcessor;
 307                                         if(i == processors - 1){
 308                                                 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
 309                                         }
 310                                         lines.push_back(linePair(startPos, numPairsPerProcessor));
 311                                 }
 312
 313                                 data = createProcesses(t, namesOfGroupCombos, true);
 314
 315                                 lines.clear();
 316                         }
 317                 #else
 318                         data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
 319                 #endif
 320
 321                 return data;
 322         }
 323         catch(exception& e) {
 324                 m->errorOut(e, "Unweighted", "getValues");
 325                 exit(1);
 326         }
 327 }
 328 /**************************************************************************************************/
 329
 330 EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, bool usingGroups) {
 331         try {
 332 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 333                 int process = 1;
 334                 int num = 0;
 335                 vector<int> processIDS;
 336
 337                 EstOutput results;
 338
 339                 //loop through and create all the processes you want
 340                 while (process != processors) {
 341                         int pid = fork();
 342
 343                         if (pid > 0) {
 344                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 345                                 process++;
 346                         }else if (pid == 0){
 347                                 EstOutput myresults;
 348                                 myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, usingGroups);
 349
 350                                 if (m->control_pressed) { exit(0); }
 351
 352                                 //pass numSeqs to parent
 353                                 ofstream out;
 354                                 string tempFile = outputDir + toString(getpid()) + ".unweighted.results.temp";
 355                                 m->openOutputFile(tempFile, out);
 356                                 out << myresults.size() << endl;
 357                                 for (int i = 0; i < myresults.size(); i++) {  out << myresults[i] << '\t';  } out << endl;
 358                                 out.close();
 359
 360                                 exit(0);
 361                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 362                 }
 363
 364                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, usingGroups);
 365
 366                 //force parent to wait until all the processes are done
 367                 for (int i=0;i<(processors-1);i++) {
 368                         int temp = processIDS[i];
 369                         wait(&temp);
 370                 }
 371
 372                 if (m->control_pressed) { return results; }
 373
 374                 //get data created by processes
 375                 for (int i=0;i<(processors-1);i++) {
 376                         ifstream in;
 377                         string s = outputDir + toString(processIDS[i]) + ".unweighted.results.temp";
 378                         m->openInputFile(s, in);
 379
 380                         //get quantiles
 381                         if (!in.eof()) {
 382                                 int num;
 383                                 in >> num; m->gobble(in);
 384
 385                                 if (m->control_pressed) { break; }
 386
 387                                 double w;
 388                                 for (int j = 0; j < num; j++) {
 389                                         in >> w;
 390                                         results.push_back(w);
 391                                 }
 392                                 m->gobble(in);
 393                         }
 394                         in.close();
 395                         remove(s.c_str());
 396                 }
 397
 398                 return results;
 399 #endif
 400         }
 401         catch(exception& e) {
 402                 m->errorOut(e, "Unweighted", "createProcesses");
 403                 exit(1);
 404         }
 405 }
 406 /**************************************************************************************************/
 407 EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, bool usingGroups) {
 408  try {
 409
 410                 EstOutput results; results.resize(num);
 411
 412                 int count = 0;
 413                 int total = num;
 414                 int twentyPercent = (total * 0.20);
 415                 int numLeaves = t->getNumLeaves();
 416
 417                 Tree* copyTree = new Tree;
 418
 419                 for (int h = start; h < (start+num); h++) {
 420
 421                         if (m->control_pressed) { return results; }
 422
 423                         //copy random tree passed in
 424                         copyTree->getCopy(t);
 425
 426                         //swap labels in the groups you want to compare
 427                         copyTree->assembleRandomUnifracTree(namesOfGroupCombos[h]);
 428
 429                         double UniqueBL=0.0000;  //a branch length is unique if it's chidren are from the same group
 430                         double totalBL = 0.00;  //all branch lengths
 431                         double UW = 0.00;               //Unweighted Value = UniqueBL / totalBL;
 432                         map<int, double> tempTotals; //maps node to total Branch Length
 433                         map<int, int> nodePcountSize; //maps node to pcountSize
 434
 435                         for(int i=0;i<copyTree->getNumNodes();i++){
 436
 437                                 if (m->control_pressed) {  return data; }
 438
 439                                 //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
 440                                 //pcountSize = 2, not unique to one group
 441                                 //pcountSize = 1, unique to one group
 442
 443                                 int pcountSize = 0;
 444                                 for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
 445                                         map<string, int>::iterator itGroup = copyTree->tree[i].pcount.find(namesOfGroupCombos[h][j]);
 446                                         if (itGroup != copyTree->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } }
 447                                 }
 448
 449                                 nodePcountSize[i] = pcountSize;
 450
 451                                 if (pcountSize == 0) { }
 452                                 else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize == 1)) {  UniqueBL += abs(copyTree->tree[i].getBranchLength());     }
 453
 454                                 //if you are a leaf from a users group add to total
 455                                 if (i < numLeaves) {
 456                                         if ((copyTree->tree[i].getBranchLength() != -1) && pcountSize != 0) {
 457                                                 totalBL += abs(copyTree->tree[i].getBranchLength());
 458                                         }
 459                                         tempTotals[i] = 0.0;  //we don't care about you, or we have already added you
 460                                 }else{ //if you are not a leaf
 461                                         //do both your chidren have have descendants from the users groups?
 462                                         int lc = copyTree->tree[i].getLChild();
 463                                         int rc = copyTree->tree[i].getRChild();
 464
 465                                         //if yes, add to total your childrens branchLengths and your childrens tempTotals
 466                                         if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
 467                                                 totalBL += tempTotals[lc] + tempTotals[rc];
 468                                                 if (copyTree->tree[i].getBranchLength() != -1) {
 469                                                         tempTotals[i] = abs(copyTree->tree[i].getBranchLength());
 470                                                 }else {
 471                                                         tempTotals[i] = 0.0;
 472                                                 }
 473                                                 //tempTotals[i] = tempTotals[lc] + tempTotals[rc];
 474                                         }else { //if no, your tempTotal is your childrens temp totals
 475                                                 tempTotals[i] = tempTotals[lc] + tempTotals[rc] + abs(copyTree->tree[i].getBranchLength());
 476                                         }
 477                                 }
 478
 479                         }
 480
 481                         UW = (UniqueBL / totalBL);
 482
 483                         if (isnan(UW) || isinf(UW)) { UW = 0; }
 484
 485                         results[count] = UW;
 486                         count++;
 487
 488                 }
 489
 490                 delete copyTree;
 491
 492                 return results;
 493         }
 494         catch(exception& e) {
 495                 m->errorOut(e, "Unweighted", "driver");
 496                 exit(1);
 497         }
 498 }
 499 /**************************************************************************************************/
 500
 501