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