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