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