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
  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                                         //require at least 4 SNPs on each side of the break
  50                                         if ((numSNPSLeft >= 4) && (numSNPSRight >= 4)) {
  51
  52                                                 float BS_A, BS_B;
  53                                                 bootstrapSNPS(snpsLeft, snpsRight, BS_A, BS_B, iters);
  54
  55                                                 if (m->control_pressed) { delete q; delete leftParent; delete rightParent; return "no"; }
  56
  57                                                 divs[k].bsa = BS_A;
  58                                                 divs[k].bsb = BS_B;
  59                                                 divs[k].bsMax = max(BS_A, BS_B);
  60                                                 divs[k].chimeraMax = max(divs[k].qla_qrb, divs[k].qlb_qra);
  61
  62
  63                                                 //are we within 10 points of the bootstrap cutoff?
  64                                                 if ((divs[k].bsMax >= (minBS-10)) && (iters < 1000)) {
  65                                                         bootstrapSNPS(snpsLeft, snpsRight, BS_A, BS_B, 1000);
  66
  67                                                         if (m->control_pressed) { delete q; delete leftParent; delete rightParent; return "no"; }
  68
  69                                                         divs[k].bsa = BS_A;
  70                                                         divs[k].bsb = BS_B;
  71                                                         divs[k].bsMax = max(BS_A, BS_B);
  72                                                         divs[k].chimeraMax = max(divs[k].qla_qrb, divs[k].qlb_qra);
  73                                                 }
  74
  75                                                 //so results reflect orignal alignment
  76                                                 divs[k].winLStart = spots[divs[k].winLStart];
  77                                                 divs[k].winLEnd = spots[divs[k].winLEnd];
  78                                                 divs[k].winRStart = spots[divs[k].winRStart];
  79                                                 divs[k].winREnd = spots[divs[k].winREnd];
  80
  81                                                 selectedDivs.push_back(divs[k]);
  82                                         }
  83                                 }
  84
  85                                 //save selected
  86                                 for (int mi = 0; mi < selectedDivs.size(); mi++) {  all.push_back(selectedDivs[mi]);    }
  87
  88                                 delete q;
  89                                 delete leftParent;
  90                                 delete rightParent;
  91                         }
  92                 }
  93
  94
  95                 // compute bootstrap support
  96                 if (all.size() > 0) {
  97                         //sort them
  98                         sort(all.begin(), all.end(), compareDataStruct);
  99                         reverse(all.begin(), all.end());
 100
 101                         outputResults = all;
 102                         return "yes";
 103                 }else {
 104                         outputResults = all;
 105                         return "no";
 106                 }
 107         }
 108         catch(exception& e) {
 109                 m->errorOut(e, "Slayer", "getResults");
 110                 exit(1);
 111         }
 112 }
 113 /***********************************************************************/
 114 vector<data_struct> Slayer::runBellerophon(Sequence* q, Sequence* pA, Sequence* pB, map<int, int>& spots) {
 115         try{
 116
 117                 vector<data_struct> data;
 118
 119                 //vertical filter
 120                 vector<Sequence*> temp;
 121                 temp.push_back(q); temp.push_back(pA); temp.push_back(pB);
 122
 123                 //maps spot in new alignment to spot in alignment before filter
 124                 spots = verticalFilter(temp);  //fills baseSpots
 125
 126                 //get these to avoid numerous function calls
 127                 string query = q->getAligned();
 128                 string parentA = pA->getAligned();
 129                 string parentB = pB->getAligned();
 130                 int length = query.length();
 131 //cout << q->getName() << endl << q->getAligned() << endl << endl;
 132 //cout << pA->getName() << endl << pA->getUnaligned() << endl << endl;
 133 //cout << pB->getName() << endl << pB->getUnaligned() << endl << endl;
 134 //cout << " length = " << length << endl;
 135
 136                 //check window size
 137                 if (length < (2*windowSize+windowStep)) {
 138 //                      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();
 139                         windowSize = length / 4;
 140                 }
 141
 142                 for (int i = windowSize-1; i <= (length - windowSize); i += windowStep) {
 143
 144                         if (m->control_pressed) { return data; }
 145
 146                         int breakpoint = i;
 147                         int leftLength = breakpoint + 1;
 148                         int rightLength = length - leftLength;
 149
 150                         float QLA = computePercentID(query, parentA, 0, breakpoint);
 151                         float QRB = computePercentID(query, parentB, breakpoint+1, length);
 152
 153                         float QLB = computePercentID(query, parentB, 0, breakpoint);
 154                         float QRA = computePercentID(query, parentA, breakpoint+1, length);
 155
 156                         float LAB = computePercentID(parentA, parentB, 0, breakpoint);
 157                         float RAB = computePercentID(parentA, parentB, breakpoint+1, length);
 158
 159                         float AB = ((LAB*leftLength) + (RAB*rightLength)) / (float) length;
 160                         float QA = ((QLA*leftLength) + (QRA*rightLength)) / (float) length;
 161                         float QB = ((QLB*leftLength) + (QRB*rightLength)) / (float) length;
 162
 163                         float QLA_QRB = ((QLA*leftLength) + (QRB*rightLength)) / (float) length;
 164                         float QLB_QRA = ((QLB*leftLength) + (QRA*rightLength)) / (float) length;
 165
 166                         //in original and not used
 167                         //float avgQA_QB = ((QA*leftLength) + (QB*rightLength)) / (float) length;
 168
 169                         float divR_QLA_QRB = min((QLA_QRB/QA), (QLA_QRB/QB));
 170                         float divR_QLB_QRA = min((QLB_QRA/QA), (QLB_QRA/QB));
 171
 172
 173                         //cout << q->getName() << '\t';
 174                         //cout << pA->getName() << '\t';
 175                         //cout << pB->getName() << '\t';
 176                    // cout << "bp: " << breakpoint << " CHIM_TYPE_A\t" << divR_QLA_QRB << "\tQLA: " << QLA << "\tQRB: " << QRB << "\tQLA_QRB: " << QLA_QRB;
 177                         //cout << "\tCHIM_TYPE_B\t" << divR_QLB_QRA << "\tQLB: " << QLB << "\tQRA: " << QRA << "\tQLB_QRA: " << QLB_QRA << endl;
 178 //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;
 179
 180 //cout << divRThreshold << endl;
 181 //cout << breakpoint << '\t' << divR_QLA_QRB << '\t' << divR_QLB_QRA << endl;
 182                         //is one of them above the
 183                         if (divR_QLA_QRB >= divRThreshold || divR_QLB_QRA >= divRThreshold) {
 184
 185                                 if (((QLA_QRB > QA) && (QLA_QRB > QB) && (QLA >= parentFragmentThreshold) && (QRB >= parentFragmentThreshold))  ||
 186                                         ((QLB_QRA > QA) && (QLB_QRA > QB) && (QLB >=parentFragmentThreshold) && (QRA >= parentFragmentThreshold)))  {
 187
 188                                         data_struct member;
 189
 190                                         member.divr_qla_qrb = divR_QLA_QRB;
 191                                         member.divr_qlb_qra = divR_QLB_QRA;
 192                                         member.qla_qrb = QLA_QRB;
 193                                         member.qlb_qra = QLB_QRA;
 194                                         member.qla = QLA;
 195                                         member.qrb = QRB;
 196                                         member.ab = AB;
 197                                         member.qa = QA;
 198                                         member.qb = QB;
 199                                         member.lab = LAB;
 200                                         member.rab = RAB;
 201                                         member.qra = QRA;
 202                                         member.qlb = QLB;
 203                                         member.winLStart = 0;
 204                                         member.winLEnd = breakpoint;
 205                                         member.winRStart = breakpoint+1;
 206                                         member.winREnd = length-1;
 207                                         member.querySeq = *(q);
 208                                         member.parentA = *(pA);
 209                                         member.parentB = *(pB);
 210                                         member.bsa = 0;
 211                                         member.bsb = 0;
 212                                         member.bsMax = 0;
 213                                         member.chimeraMax = 0;
 214
 215                                         data.push_back(member);
 216
 217                                 }//if
 218                         }//if
 219                 }//for
 220
 221
 222                 return data;
 223
 224         }
 225         catch(exception& e) {
 226                 m->errorOut(e, "Slayer", "runBellerophon");
 227                 exit(1);
 228         }
 229 }
 230 /***********************************************************************/
 231 vector<snps> Slayer::getSNPS(string parentA, string query, string parentB, int left, int right) {
 232         try {
 233
 234                 vector<snps> data;
 235
 236                 for (int i = left; i <= right; i++) {
 237
 238                         char A = parentA[i];
 239                         char Q = query[i];
 240                         char B = parentB[i];
 241
 242                         if ((A != Q) || (B != Q)) {
 243
 244                                 //ensure not neighboring a gap. change to 12/09 release of chimeraSlayer - not sure what this adds, but it eliminates alot of SNPS
 245
 246
 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()-1 || 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()-1 || 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()-1 || abs (baseSpots[2][i] - baseSpots[2][i+1]) == 1)
 259                                         )
 260                                 {
 261                                         snps member;
 262                                         member.queryChar = Q;
 263                                         member.parentAChar = A;
 264                                         member.parentBChar = B;
 265                                         data.push_back(member);
 266                                 }
 267                         }
 268                 }
 269
 270                 return data;
 271
 272         }
 273         catch(exception& e) {
 274                 m->errorOut(e, "Slayer", "getSNPS");
 275                 exit(1);
 276         }
 277 }
 278 /***********************************************************************/
 279 int Slayer::bootstrapSNPS(vector<snps> left, vector<snps> right, float& BSA, float& BSB, int numIters) {
 280         try {
 281
 282                 srand((unsigned)time( NULL ));
 283
 284                 int count_A = 0; // sceneario QLA,QRB supported
 285                 int count_B = 0; // sceneario QLB,QRA supported
 286
 287                 int numLeft = max(1, int(left.size() * percentSNPSample/(float)100 + 0.5));
 288                 int numRight = max(1, int(right.size() * percentSNPSample/(float)100 + 0.5));
 289
 290                 for (int i = 0; i < numIters; i++) {
 291                         //random sampling with replacement.
 292
 293                         if (m->control_pressed) { return 0;  }
 294
 295                         vector<snps> selectedLeft;
 296
 297                         for (int j = 0; j < numLeft; j++) {
 298                                 int index = int(rand() % left.size());
 299                                 selectedLeft.push_back(left[index]);
 300                         }
 301
 302                         vector<snps> selectedRight;
 303                         for (int j = 0; j < numRight; j++) {
 304                                 int index = int(rand() % right.size());
 305                                 selectedRight.push_back(right[index]);
 306                         }
 307
 308                         /* A  ------------------------------------------
 309                         #       QLA                     QRA
 310                         # Q  ------------------------------------------
 311                         #                      |
 312                         #                      |
 313                         # Q  ------------------------------------------
 314                         #       QLB                     QRB
 315                         # B  ------------------------------------------ */
 316
 317
 318                         float QLA = snpQA(selectedLeft);
 319                         float QRA = snpQA(selectedRight);
 320
 321                         float QLB = snpQB(selectedLeft);
 322                         float QRB = snpQB(selectedRight);
 323
 324                         //in original - not used - not sure why?
 325                         //float ALB = snpAB(selectedLeft);
 326                         //float ARB = snpAB(selectedRight);
 327
 328                         if ((QLA > QLB) && (QRB > QRA)) {
 329                                 count_A++;
 330                         }
 331
 332                         if ((QLB > QLA) && (QRA > QRB)) {
 333                                 count_B++;
 334                         }
 335
 336 //cout << "selected left snp: \n";
 337 //for (int j = 0; j < selectedLeft.size(); j++) {  cout << selectedLeft[j].parentAChar;  }
 338 //cout << endl;
 339 //for (int j = 0; j < selectedLeft.size(); j++) {  cout << selectedLeft[j].queryChar;  }
 340 //cout << endl;
 341 //for (int j = 0; j < selectedLeft.size(); j++) {  cout << selectedLeft[j].parentBChar;  }
 342 //cout << endl;
 343 //cout << "selected right snp: \n";
 344 //for (int j = 0; j < selectedRight.size(); j++) {  cout << selectedRight[j].parentAChar;  }
 345 //cout << endl;
 346 //for (int i = 0; i < selectedRight.size(); i++) {  cout << selectedRight[i].queryChar;  }
 347 //cout << endl;
 348 //for (int i = 0; i < selectedRight.size(); i++) {  cout << selectedRight[i].parentBChar;  }
 349 //cout << endl;
 350                 }
 351
 352
 353                 //cout << count_A << '\t' << count_B << endl;
 354
 355                 BSA = (float) count_A / (float) numIters * 100;
 356                 BSB = (float) count_B / (float) numIters * 100;
 357 //cout << "bsa = " << BSA << " bsb = " << BSB << endl;
 358
 359                 return 0;
 360
 361         }
 362         catch(exception& e) {
 363                 m->errorOut(e, "Slayer", "bootstrapSNPS");
 364                 exit(1);
 365         }
 366 }
 367 /***********************************************************************/
 368 float Slayer::snpQA(vector<snps> data) {
 369         try {
 370
 371                 int numIdentical = 0;
 372
 373                 for (int i = 0; i < data.size(); i++) {
 374                         if (data[i].parentAChar == data[i].queryChar) {
 375                                 numIdentical++;
 376                         }
 377                 }
 378
 379                 float percentID = (numIdentical / (float) data.size()) * 100;
 380
 381                 return percentID;
 382         }
 383         catch(exception& e) {
 384                 m->errorOut(e, "Slayer", "snpQA");
 385                 exit(1);
 386         }
 387 }
 388 /***********************************************************************/
 389 float Slayer::snpQB(vector<snps> data) {
 390         try {
 391
 392                 int numIdentical = 0;
 393
 394                 for (int i = 0; i < data.size(); i++) {
 395                         if (data[i].parentBChar == data[i].queryChar) {
 396                                 numIdentical++;
 397                         }
 398                 }
 399
 400                 float percentID = (numIdentical / (float) data.size()) * 100;
 401
 402                 return percentID;
 403
 404         }
 405         catch(exception& e) {
 406                 m->errorOut(e, "Slayer", "snpQB");
 407                 exit(1);
 408         }
 409 }
 410 /***********************************************************************/
 411 float Slayer::snpAB(vector<snps> data) {
 412         try {
 413                 int numIdentical = 0;
 414
 415                 for (int i = 0; i < data.size(); i++) {
 416                         if (data[i].parentAChar == data[i].parentBChar) {
 417                                 numIdentical++;
 418                         }
 419                 }
 420
 421                 float percentID = (numIdentical / (float) data.size()) * 100;
 422
 423                 return percentID;
 424
 425         }
 426         catch(exception& e) {
 427                 m->errorOut(e, "Slayer", "snpAB");
 428                 exit(1);
 429         }
 430 }
 431 /***********************************************************************/
 432 float Slayer::computePercentID(string queryAlign, string chimera, int left, int right) {
 433         try {
 434
 435                 int numIdentical = 0;
 436                 int countA = 0;
 437                 int countB = 0;
 438                 for (int i = left; i <= right; i++) {
 439                         if (((queryAlign[i] != 'G') && (queryAlign[i] != 'T') && (queryAlign[i] != 'A') && (queryAlign[i] != 'C')&& (queryAlign[i] != '.') && (queryAlign[i] != '-')) ||
 440                                 ((chimera[i] != 'G') && (chimera[i] != 'T') && (chimera[i] != 'A') && (chimera[i] != 'C')&& (chimera[i] != '.') && (chimera[i] != '-'))) {}
 441                         else {
 442
 443                                 bool charA = false; bool charB = false;
 444                                 if ((queryAlign[i] == 'G') || (queryAlign[i] == 'T') || (queryAlign[i] == 'A') || (queryAlign[i] == 'C')) { charA = true; }
 445                                 if ((chimera[i] == 'G') || (chimera[i] == 'T') || (chimera[i] == 'A') || (chimera[i] == 'C')) { charB = true; }
 446
 447                                 if (charA || charB) {
 448
 449                                         if (charA) { countA++; }
 450                                         if (charB) { countB++; }
 451
 452                                         if (queryAlign[i] == chimera[i]) {
 453                                                 numIdentical++;
 454                                         }
 455                                 }
 456                         }
 457
 458                 }
 459
 460                 float numBases = (countA + countB) /(float) 2;
 461
 462                 if (numBases == 0) { return 0; }
 463
 464                 float percentIdentical = (numIdentical/(float)numBases) * 100;
 465
 466                 return percentIdentical;
 467
 468         }
 469         catch(exception& e) {
 470                 m->errorOut(e, "Slayer", "computePercentID");
 471                 exit(1);
 472         }
 473 }
 474 /***********************************************************************/
 475 //remove columns that contain any gaps
 476 map<int, int> Slayer::verticalFilter(vector<Sequence*> seqs) {
 477         try {
 478                 //find baseSpots
 479                 baseSpots.clear();
 480                 baseSpots.resize(3);  //query, parentA, parentB
 481
 482                 vector<int> gaps;       gaps.resize(seqs[0]->getAligned().length(), 0);
 483
 484                 string filterString = (string(seqs[0]->getAligned().length(), '1'));
 485
 486                 //for each sequence
 487                 for (int i = 0; i < seqs.size(); i++) {
 488
 489                         string seqAligned = seqs[i]->getAligned();
 490
 491                         for (int j = 0; j < seqAligned.length(); j++) {
 492                                 //if this spot is a gap
 493                                 if ((seqAligned[j] == '-') || (seqAligned[j] == '.') || (toupper(seqAligned[j]) == 'N'))        {   gaps[j]++;  }
 494                         }
 495                 }
 496
 497                 //zero out spot where any sequences have blanks
 498                 int numColRemoved = 0;
 499                 int count = 0;
 500                 map<int, int> maskMap; maskMap.clear();
 501
 502                 for(int i = 0; i < seqs[0]->getAligned().length(); i++){
 503                         if(gaps[i] != 0)        {       filterString[i] = '0';  numColRemoved++;  }
 504                         else {
 505                                 maskMap[count] = i;
 506                                 count++;
 507                         }
 508                 }
 509
 510                 //for each sequence
 511                 for (int i = 0; i < seqs.size(); i++) {
 512
 513                         string seqAligned = seqs[i]->getAligned();
 514                         string newAligned = "";
 515
 516                         int baseCount = 0;
 517                         int count = 0;
 518                         for (int j = 0; j < seqAligned.length(); j++) {
 519                                 //are you a base
 520                                 if ((seqAligned[j] != '-') && (seqAligned[j] != '.') && (toupper(seqAligned[j]) != 'N'))        { baseCount++; }
 521
 522                                 //if this spot is not a gap
 523                                 if (filterString[j] == '1') {
 524                                         newAligned += seqAligned[j];
 525                                         baseSpots[i][count] = baseCount;
 526                                         count++;
 527                                 }
 528                         }
 529
 530                         seqs[i]->setAligned(newAligned);
 531                 }
 532
 533                 return maskMap;
 534         }
 535         catch(exception& e) {
 536                 m->errorOut(e, "Slayer", "verticalFilter");
 537                 exit(1);
 538         }
 539 }
 540 /***********************************************************************/