slayer.cpp

   1 /*
   2  *  slayer.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/25/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "slayer.h"
  11
  12 /***********************************************************************/
  13 Slayer::Slayer(int win, int increment, int parentThreshold, float div, int i, int snp, int mi) :
  14                 minBS(mi), windowSize(win), windowStep(increment), parentFragmentThreshold(parentThreshold), divRThreshold(div), iters(i), percentSNPSample(snp){ m = MothurOut::getInstance(); }
  15 /***********************************************************************/
  16 string Slayer::getResults(Sequence* query, vector<Sequence*> refSeqs) {
  17         try {
  18                 vector<data_struct> all; all.clear();
  19                 myQuery = *query;
  20
  21                 for (int i = 0; i < refSeqs.size(); i++) {
  22
  23                         for (int j = i+1; j < refSeqs.size(); j++) {
  24
  25                                 if (m->control_pressed) { return "no";  }
  26
  27                                 //make copies of query and each parent because runBellerophon removes gaps and messes them up
  28                                 Sequence* q = new Sequence(query->getName(), query->getAligned());
  29                                 Sequence* leftParent = new Sequence(refSeqs[i]->getName(), refSeqs[i]->getAligned());
  30                                 Sequence* rightParent = new Sequence(refSeqs[j]->getName(), refSeqs[j]->getAligned());
  31
  32                                 map<int, int> spots;  //map from spot in original sequence to spot in filtered sequence for query and both parents
  33                                 vector<data_struct> divs = runBellerophon(q, leftParent, rightParent, spots);
  34                                 //cout << divs.size() << endl;
  35                                 if (m->control_pressed) {
  36                                         delete q;
  37                                         delete leftParent;
  38                                         delete rightParent;
  39                                         return "no";
  40                                 }
  41
  42 //                              cout << divs.size() << endl;
  43                                 vector<data_struct> selectedDivs;
  44                                 for (int k = 0; k < divs.size(); k++) {
  45
  46                                         vector<snps> snpsLeft = getSNPS(divs[k].parentA.getAligned(), divs[k].querySeq.getAligned(), divs[k].parentB.getAligned(), divs[k].winLStart, divs[k].winLEnd);
  47                                         vector<snps> snpsRight = getSNPS(divs[k].parentA.getAligned(), divs[k].querySeq.getAligned(), divs[k].parentB.getAligned(), divs[k].winRStart, divs[k].winREnd);
  48
  49                                         if (m->control_pressed) {
  50                                                 delete q;
  51                                                 delete leftParent;
  52                                                 delete rightParent;
  53                                                 return "no";
  54                                         }
  55
  56                                         int numSNPSLeft = snpsLeft.size();
  57                                         int numSNPSRight = snpsRight.size();
  58
  59                                         //require at least 3 SNPs on each side of the break
  60 //                                      if ((numSNPSLeft >= 3) && (numSNPSRight >= 3)) {
  61                                         if ((numSNPSLeft >= 4) && (numSNPSRight >= 4)) {
  62
  63                                                 //removed in 12/09 version of chimeraSlayer
  64                                                 //int winSizeLeft = divs[k].winLEnd - divs[k].winLStart + 1;
  65                                                 //int winSizeRight = divs[k].winREnd - divs[k].winRStart + 1;
  66
  67                                                 //float snpRateLeft = numSNPSLeft / (float) winSizeLeft;
  68                                                 //float snpRateRight = numSNPSRight / (float) winSizeRight;
  69                                                 //float logR = log(snpRateLeft / snpRateRight) / log(2.0);
  70
  71                                                 // do not accept excess snp ratio on either side of the break
  72                                                 //if (abs(logR) < 1 ) {
  73
  74                                                         float BS_A, BS_B;
  75                                                         bootstrapSNPS(snpsLeft, snpsRight, BS_A, BS_B, iters);
  76
  77                                                         if (m->control_pressed) {
  78                                                                 delete q;
  79                                                                 delete leftParent;
  80                                                                 delete rightParent;
  81                                                                 return "no";
  82                                                         }
  83
  84                                                         divs[k].bsa = BS_A;
  85                                                         divs[k].bsb = BS_B;
  86
  87                                                         divs[k].bsMax = max(BS_A, BS_B);
  88
  89                                                         divs[k].chimeraMax = max(divs[k].qla_qrb, divs[k].qlb_qra);
  90
  91                                                         //so results reflect orignal alignment
  92                                                         divs[k].winLStart = spots[divs[k].winLStart];
  93                                                         divs[k].winLEnd = spots[divs[k].winLEnd];
  94                                                         divs[k].winRStart = spots[divs[k].winRStart];
  95                                                         divs[k].winREnd = spots[divs[k].winREnd];
  96
  97                                                         selectedDivs.push_back(divs[k]);
  98                                                 //}
  99                                         }
 100                                 }
 101
 102                                 //save selected
 103                                 for (int mi = 0; mi < selectedDivs.size(); mi++) {  all.push_back(selectedDivs[mi]);    }
 104
 105                                 delete q;
 106                                 delete leftParent;
 107                                 delete rightParent;
 108                         }
 109                 }
 110
 111                 // compute bootstrap support
 112                 if (all.size() > 0) {
 113                         //sort them
 114                         sort(all.begin(), all.end(), compareDataStruct);
 115                         reverse(all.begin(), all.end());
 116                         /*vector<data_struct> newAll;
 117
 118                         for (int i = 0; i < all.size(); i++) {
 119
 120                                 //are we within 10 points of the bootstrap cutoff?
 121                                 if (all[i].bsMax >= (minBS-10)) {
 122
 123                                         vector<snps> snpsLeft = getSNPS(all[i].parentA.getAligned(), all[i].querySeq.getAligned(), all[i].parentB.getAligned(), all[i].winLStart, all[i].winLEnd);
 124                                         vector<snps> snpsRight = getSNPS(all[i].parentA.getAligned(), all[i].querySeq.getAligned(), all[i].parentB.getAligned(), all[i].winRStart, all[i].winREnd);
 125
 126                                         if (m->control_pressed) { return "no"; }
 127
 128                                         int numSNPSLeft = snpsLeft.size();
 129                                         int numSNPSRight = snpsRight.size();
 130
 131                                         //require at least 4 SNPs on each side of the break
 132                                         if ((numSNPSLeft >= 4) && (numSNPSRight >= 4)) {
 133                                                 float BS_A, BS_B;
 134                                                 bootstrapSNPS(snpsLeft, snpsRight, BS_A, BS_B, 1000);
 135
 136                                                 if (m->control_pressed) { return "no"; }
 137                 cout << i << '\t' << BS_A << '\t' << BS_B << endl;
 138                                                 all[i].bsa = BS_A;
 139                                                 all[i].bsb = BS_B;
 140                                                 all[i].bsMax = max(BS_A, BS_B);
 141                                                 all[i].chimeraMax = max(all[i].qla_qrb, all[i].qlb_qra);
 142
 143                                                 newAll.push_back(all[i]);
 144                                         }
 145                                 }
 146                         }
 147
 148                         if (newAll.size() != 0) {
 149                                 sort(newAll.begin(), newAll.end(), compareDataStruct);
 150                                 reverse(newAll.begin(), newAll.end());
 151
 152                                 outputResults = newAll;
 153                                 return "yes";
 154                         }else {
 155                                 outputResults = newAll;
 156                                 return "no";
 157                         }*/
 158
 159                         outputResults = all;
 160                         return "yes";
 161                 }
 162                 else {
 163                         outputResults = all;
 164                         return "no";
 165                 }
 166         }
 167         catch(exception& e) {
 168                 m->errorOut(e, "Slayer", "getResults");
 169                 exit(1);
 170         }
 171 }
 172 /***********************************************************************/
 173 vector<data_struct> Slayer::runBellerophon(Sequence* q, Sequence* pA, Sequence* pB, map<int, int>& spots) {
 174         try{
 175
 176                 vector<data_struct> data;
 177
 178                 //vertical filter
 179                 vector<Sequence*> temp;
 180                 temp.push_back(q); temp.push_back(pA); temp.push_back(pB);
 181
 182                 //maps spot in new alignment to spot in alignment before filter
 183                 spots = verticalFilter(temp);  //fills baseSpots
 184
 185                 //get these to avoid numerous function calls
 186                 string query = q->getAligned();
 187                 string parentA = pA->getAligned();
 188                 string parentB = pB->getAligned();
 189                 int length = query.length();
 190 /*cout << q->getName() << endl << q->getAligned() << endl << endl;
 191 cout << pA->getName() << endl << pA->getAligned() << endl << endl;
 192 cout << pB->getName() << endl << pB->getAligned() << endl << endl;
 193 cout << " length = " << length << endl;
 194 cout << q->getName() << endl;
 195 cout << pA->getName() << '\t';
 196 cout << pB->getName() << endl;*/
 197
 198                 //check window size
 199                 if (length < (2*windowSize+windowStep)) {
 200 //                      m->mothurOut("Your window size is too large for " + q->getName() + ". I will make the window size " + toString(length/4) + " which is 1/4 the filtered length."); m->mothurOutEndLine();
 201                         windowSize = length / 4;
 202                 }
 203
 204                 for (int i = windowSize-1; i <= (length - windowSize); i += windowStep) {
 205
 206                         if (m->control_pressed) { return data; }
 207
 208                         int breakpoint = i;
 209                         int leftLength = breakpoint + 1;
 210                         int rightLength = length - leftLength;
 211
 212                         float QLA = computePercentID(query, parentA, 0, breakpoint);
 213                         float QRB = computePercentID(query, parentB, breakpoint+1, length - 1);
 214
 215                         float QLB = computePercentID(query, parentB, 0, breakpoint);
 216                         float QRA = computePercentID(query, parentA, breakpoint+1, length - 1);
 217
 218                         float LAB = computePercentID(parentA, parentB, 0, breakpoint);
 219                         float RAB = computePercentID(parentA, parentB, breakpoint+1, length - 1);
 220
 221                         float AB = ((LAB*leftLength) + (RAB*rightLength)) / (float) length;
 222                         float QA = ((QLA*leftLength) + (QRA*rightLength)) / (float) length;
 223                         float QB = ((QLB*leftLength) + (QRB*rightLength)) / (float) length;
 224
 225                         float QLA_QRB = ((QLA*leftLength) + (QRB*rightLength)) / (float) length;
 226                         float QLB_QRA = ((QLB*leftLength) + (QRA*rightLength)) / (float) length;
 227
 228                         //in original and not used
 229                         //float avgQA_QB = ((QA*leftLength) + (QB*rightLength)) / (float) length;
 230
 231                         float divR_QLA_QRB = min((QLA_QRB/QA), (QLA_QRB/QB));
 232                         float divR_QLB_QRA = min((QLB_QRA/QA), (QLB_QRA/QB));
 233 //cout << leftLength << '\t' << rightLength << '\t' << QLA << '\t' << QRB << '\t' << QLB << '\t' << QRA  << '\t' << LAB << '\t' << RAB << '\t' << AB << '\t' << QA << '\t' << QB << '\t' << QLA_QRB << '\t' <<  QLB_QRA <<    endl;
 234
 235 //cout << divRThreshold << endl;
 236 //cout << breakpoint << '\t' << divR_QLA_QRB << '\t' << divR_QLB_QRA << endl;
 237                         //is one of them above the
 238                         if (divR_QLA_QRB >= divRThreshold || divR_QLB_QRA >= divRThreshold) {
 239
 240                                 if (((QLA_QRB > QA) && (QLA_QRB > QB) && (QLA >= parentFragmentThreshold) && (QRB >= parentFragmentThreshold))  ||
 241                                         ((QLB_QRA > QA) && (QLB_QRA > QB) && (QLB >=parentFragmentThreshold) && (QRA >= parentFragmentThreshold)))  {
 242
 243                                         data_struct member;
 244
 245                                         member.divr_qla_qrb = divR_QLA_QRB;
 246                                         member.divr_qlb_qra = divR_QLB_QRA;
 247                                         member.qla_qrb = QLA_QRB;
 248                                         member.qlb_qra = QLB_QRA;
 249                                         member.qla = QLA;
 250                                         member.qrb = QRB;
 251                                         member.ab = AB;
 252                                         member.qa = QA;
 253                                         member.qb = QB;
 254                                         member.lab = LAB;
 255                                         member.rab = RAB;
 256                                         member.qra = QRA;
 257                                         member.qlb = QLB;
 258                                         member.winLStart = 0;
 259                                         member.winLEnd = breakpoint;
 260                                         member.winRStart = breakpoint+1;
 261                                         member.winREnd = length-1;
 262                                         member.querySeq = *(q);
 263                                         member.parentA = *(pA);
 264                                         member.parentB = *(pB);
 265                                         member.bsa = 0;
 266                                         member.bsb = 0;
 267                                         member.bsMax = 0;
 268                                         member.chimeraMax = 0;
 269
 270                                         data.push_back(member);
 271
 272                                 }//if
 273                         }//if
 274                 }//for
 275
 276                 return data;
 277
 278         }
 279         catch(exception& e) {
 280                 m->errorOut(e, "Slayer", "runBellerophon");
 281                 exit(1);
 282         }
 283 }
 284 /***********************************************************************/
 285 vector<snps> Slayer::getSNPS(string parentA, string query, string parentB, int left, int right) {
 286         try {
 287
 288                 vector<snps> data;
 289 //cout << left << '\t' << right << endl;
 290                 for (int i = left; i <= right; i++) {
 291
 292                         char A = parentA[i];
 293                         char Q = query[i];
 294                         char B = parentB[i];
 295
 296                         if ((A != Q) || (B != Q)) {
 297 //cout << "not equal " << Q << '\t' << A << '\t' << B << endl;
 298
 299                                 //ensure not neighboring a gap. change to 12/09 release of chimeraSlayer - not sure what this adds, but it eliminates alot of SNPS
 300                                 if (
 301                                         //did query loose a base here during filter??
 302                                         ( i == 0 || abs (baseSpots[0][i] - baseSpots[0][i-1]) == 1) &&
 303                                         ( i == query.length() || abs (baseSpots[0][i] - baseSpots[0][i+1]) == 1)
 304                                         &&
 305                                         //did parentA loose a base here during filter??
 306                                         ( i == 0 || abs (baseSpots[1][i] - baseSpots[1][i-1]) == 1) &&
 307                                         ( i == parentA.length() || abs (baseSpots[1][i] - baseSpots[1][i+1]) == 1)
 308                                         &&
 309                                         //did parentB loose a base here during filter??
 310                                         ( i == 0 || abs (baseSpots[2][i] - baseSpots[2][i-1]) == 1) &&
 311                                         ( i == parentB.length() || abs (baseSpots[2][i] - baseSpots[2][i+1]) == 1)
 312                                         )
 313                                 {
 314
 315                                         snps member;
 316                                         member.queryChar = Q;
 317                                         member.parentAChar = A;
 318                                         member.parentBChar = B;
 319 //cout << "not neighboring a gap " << Q << '\t' << A << '\t' << B << '\t' << baseSpots[0][i] << '\t' << baseSpots[0][i+1] << '\t' << baseSpots[0][i-1] << '\t' << baseSpots[1][i] << '\t' << baseSpots[1][i+1] << '\t' << baseSpots[1][i-1] << '\t' << baseSpots[2][i] << '\t' << baseSpots[2][i+1] << '\t' << baseSpots[2][i-1] << endl;
 320                                         data.push_back(member);
 321                                 }
 322                         }
 323 //                      cout << i << '\t' << data.size() << endl;
 324                 }
 325
 326                 return data;
 327
 328         }
 329         catch(exception& e) {
 330                 m->errorOut(e, "Slayer", "getSNPS");
 331                 exit(1);
 332         }
 333 }
 334 /***********************************************************************/
 335 int Slayer::bootstrapSNPS(vector<snps> left, vector<snps> right, float& BSA, float& BSB, int numIters) {
 336         try {
 337
 338                 srand((unsigned)time( NULL ));
 339
 340                 int count_A = 0; // sceneario QLA,QRB supported
 341                 int count_B = 0; // sceneario QLB,QRA supported
 342
 343                 int numLeft = max(1, int(left.size() * percentSNPSample/(float)100 + 0.5));
 344                 int numRight = max(1, int(right.size() * percentSNPSample/(float)100 + 0.5));
 345
 346                 for (int i = 0; i < numIters; i++) {
 347                         //random sampling with replacement.
 348
 349                         if (m->control_pressed) { return 0;  }
 350
 351                         vector<snps> selectedLeft;
 352
 353                         for (int j = 0; j < numLeft; j++) {
 354                                 int index = int(rand() % left.size());
 355                                 selectedLeft.push_back(left[index]);
 356                         }
 357
 358                         vector<snps> selectedRight;
 359                         for (int j = 0; j < numRight; j++) {
 360                                 int index = int(rand() % right.size());
 361                                 selectedRight.push_back(right[index]);
 362                         }
 363
 364                         /* A  ------------------------------------------
 365                         #       QLA                     QRA
 366                         # Q  ------------------------------------------
 367                         #                      |
 368                         #                      |
 369                         # Q  ------------------------------------------
 370                         #       QLB                     QRB
 371                         # B  ------------------------------------------ */
 372
 373
 374                         float QLA = snpQA(selectedLeft);
 375                         float QRA = snpQA(selectedRight);
 376
 377                         float QLB = snpQB(selectedLeft);
 378                         float QRB = snpQB(selectedRight);
 379
 380                         //in original - not used - not sure why?
 381                         //float ALB = snpAB(selectedLeft);
 382                         //float ARB = snpAB(selectedRight);
 383
 384                         if ((QLA > QLB) && (QRB > QRA)) {
 385                                 count_A++;
 386                         }
 387
 388                         if ((QLB > QLA) && (QRA > QRB)) {
 389                                 count_B++;
 390                         }
 391
 392 //cout << "selected left snp: \n";
 393 //for (int j = 0; j < selectedLeft.size(); j++) {  cout << selectedLeft[j].parentAChar;  }
 394 //cout << endl;
 395 //for (int j = 0; j < selectedLeft.size(); j++) {  cout << selectedLeft[j].queryChar;  }
 396 //cout << endl;
 397 //for (int j = 0; j < selectedLeft.size(); j++) {  cout << selectedLeft[j].parentBChar;  }
 398 //cout << endl;
 399 //cout << "selected right snp: \n";
 400 //for (int j = 0; j < selectedRight.size(); j++) {  cout << selectedRight[j].parentAChar;  }
 401 //cout << endl;
 402 //for (int i = 0; i < selectedRight.size(); i++) {  cout << selectedRight[i].queryChar;  }
 403 //cout << endl;
 404 //for (int i = 0; i < selectedRight.size(); i++) {  cout << selectedRight[i].parentBChar;  }
 405 //cout << endl;
 406                 }
 407
 408
 409                 //cout << count_A << '\t' << count_B << endl;
 410
 411                 BSA = (float) count_A / (float) numIters * 100;
 412                 BSB = (float) count_B / (float) numIters * 100;
 413 //cout << "bsa = " << BSA << " bsb = " << BSB << endl;
 414
 415                 //run borderline bootstrap values longer
 416                 //if (numIters < 1000) {
 417                         //are you within 10 points of min bootstrap value cutoff
 418                 //      if (((abs((double)(BSA - minBS))) <= 5) || ((abs((double)(BSB - minBS))) <= 5)) {
 419                 //              m->mothurOut("extending bootstrap for " + myQuery.getName()); m->mothurOutEndLine();
 420                 //              bootstrapSNPS(left, right, BSA, BSB, 1000);
 421                 //      }
 422                 //}
 423
 424                 return 0;
 425
 426         }
 427         catch(exception& e) {
 428                 m->errorOut(e, "Slayer", "bootstrapSNPS");
 429                 exit(1);
 430         }
 431 }
 432 /***********************************************************************/
 433 float Slayer::snpQA(vector<snps> data) {
 434         try {
 435
 436                 int numIdentical = 0;
 437
 438                 for (int i = 0; i < data.size(); i++) {
 439                         if (data[i].parentAChar == data[i].queryChar) {
 440                                 numIdentical++;
 441                         }
 442                 }
 443
 444                 float percentID = (numIdentical / (float) data.size()) * 100;
 445
 446                 return percentID;
 447         }
 448         catch(exception& e) {
 449                 m->errorOut(e, "Slayer", "snpQA");
 450                 exit(1);
 451         }
 452 }
 453 /***********************************************************************/
 454 float Slayer::snpQB(vector<snps> data) {
 455         try {
 456
 457                 int numIdentical = 0;
 458
 459                 for (int i = 0; i < data.size(); i++) {
 460                         if (data[i].parentBChar == data[i].queryChar) {
 461                                 numIdentical++;
 462                         }
 463                 }
 464
 465                 float percentID = (numIdentical / (float) data.size()) * 100;
 466
 467                 return percentID;
 468
 469         }
 470         catch(exception& e) {
 471                 m->errorOut(e, "Slayer", "snpQB");
 472                 exit(1);
 473         }
 474 }
 475 /***********************************************************************/
 476 float Slayer::snpAB(vector<snps> data) {
 477         try {
 478                 int numIdentical = 0;
 479
 480                 for (int i = 0; i < data.size(); i++) {
 481                         if (data[i].parentAChar == data[i].parentBChar) {
 482                                 numIdentical++;
 483                         }
 484                 }
 485
 486                 float percentID = (numIdentical / (float) data.size()) * 100;
 487
 488                 return percentID;
 489
 490         }
 491         catch(exception& e) {
 492                 m->errorOut(e, "Slayer", "snpAB");
 493                 exit(1);
 494         }
 495 }
 496 /***********************************************************************/
 497 float Slayer::computePercentID(string queryFrag, string parent, int left, int right) {
 498         try {
 499                 int total = 0;
 500                 int matches = 0;
 501
 502                 for (int i = left; i <= right; i++) {
 503                         total++;
 504                         if (queryFrag[i] == parent[i]) {
 505                                 matches++;
 506                         }
 507                 }
 508
 509                 float percentID =( matches/(float)total) * 100;
 510
 511                 return percentID;
 512         }
 513         catch(exception& e) {
 514                 m->errorOut(e, "Slayer", "computePercentID");
 515                 exit(1);
 516         }
 517 }
 518 /***********************************************************************/
 519 //remove columns that contain any gaps
 520 map<int, int> Slayer::verticalFilter(vector<Sequence*> seqs) {
 521         try {
 522                 //find baseSpots
 523                 baseSpots.clear();
 524                 baseSpots.resize(3);  //query, parentA, parentB
 525
 526                 vector<int> gaps;       gaps.resize(seqs[0]->getAligned().length(), 0);
 527
 528                 string filterString = (string(seqs[0]->getAligned().length(), '1'));
 529
 530                 //for each sequence
 531                 for (int i = 0; i < seqs.size(); i++) {
 532
 533                         string seqAligned = seqs[i]->getAligned();
 534
 535                         for (int j = 0; j < seqAligned.length(); j++) {
 536                                 //if this spot is a gap
 537                                 if ((seqAligned[j] == '-') || (seqAligned[j] == '.') || (toupper(seqAligned[j]) == 'N'))        {   gaps[j]++;  }
 538                         }
 539                 }
 540
 541                 //zero out spot where any sequences have blanks
 542                 int numColRemoved = 0;
 543                 int count = 0;
 544                 map<int, int> maskMap; maskMap.clear();
 545
 546                 for(int i = 0; i < seqs[0]->getAligned().length(); i++){
 547                         if(gaps[i] != 0)        {       filterString[i] = '0';  numColRemoved++;  }
 548                         else {
 549                                 maskMap[count] = i;
 550                                 count++;
 551                         }
 552                 }
 553
 554                 //for each sequence
 555                 for (int i = 0; i < seqs.size(); i++) {
 556
 557                         string seqAligned = seqs[i]->getAligned();
 558                         string newAligned = "";
 559
 560                         int baseCount = 0;
 561                         int count = 0;
 562                         for (int j = 0; j < seqAligned.length(); j++) {
 563                                 //are you a base
 564                                 if ((seqAligned[j] != '-') && (seqAligned[j] != '.') && (toupper(seqAligned[j]) != 'N'))        { baseCount++; }
 565
 566                                 //if this spot is not a gap
 567                                 if (filterString[j] == '1') {
 568                                         newAligned += seqAligned[j];
 569                                         baseSpots[i][count] = baseCount;
 570                                         count++;
 571                                 }
 572                         }
 573
 574                         seqs[i]->setAligned(newAligned);
 575                 }
 576
 577                 return maskMap;
 578         }
 579         catch(exception& e) {
 580                 m->errorOut(e, "Slayer", "verticalFilter");
 581                 exit(1);
 582         }
 583 }
 584 /***********************************************************************/