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) :
  14                 windowSize(win), windowStep(increment), parentFragmentThreshold(parentThreshold), divRThreshold(div) {}
  15 /***********************************************************************/
  16 string Slayer::getResults(Sequence* query, vector<Sequence*> refSeqs) {
  17         try {
  18 cout << "refSeqs = " << refSeqs.size() << endl;
  19                 vector<data_struct> all; all.clear();
  20
  21                 for (int i = 0; i < refSeqs.size(); i++) {
  22
  23                         for (int j = i+1; j < refSeqs.size(); j++) {
  24
  25                                 //make copies of query and each parent because runBellerophon removes gaps and messes them up
  26                                 Sequence* q = new Sequence(query->getName(), query->getAligned());
  27                                 Sequence* leftParent = new Sequence(refSeqs[i]->getName(), refSeqs[i]->getAligned());
  28                                 Sequence* rightParent = new Sequence(refSeqs[j]->getName(), refSeqs[j]->getAligned());
  29
  30                                 vector<data_struct> divs = runBellerophon(q, leftParent, rightParent);
  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                                                 int winSizeLeft = divs[k].winLEnd - divs[k].winLStart + 1;
  45                                                 int winSizeRight = divs[k].winREnd - divs[k].winRStart + 1;
  46
  47                                                 float snpRateLeft = numSNPSLeft / (float) winSizeLeft;
  48                                                 float snpRateRight = numSNPSRight / (float) winSizeRight;
  49
  50                                                 float logR = log(snpRateLeft / snpRateRight) / log(2);
  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                                                         selectedDivs.push_back(divs[k]);
  66                                                 }
  67                                         }
  68                                 }
  69
  70                                 //save selected
  71                                 for (int m = 0; m < selectedDivs.size(); m++) {  all.push_back(selectedDivs[m]);        }
  72
  73                                 delete q;
  74                                 delete leftParent;
  75                                 delete rightParent;
  76                         }
  77                 }
  78
  79
  80                 // compute bootstrap support
  81                 if (all.size() > 0) {
  82                         //sort them
  83                         sort(all.begin(), all.end(), compareDataStruct);
  84                         reverse(all.begin(), all.end());
  85
  86                         outputResults = all;
  87                         return "yes";
  88                 }
  89                 else {
  90                         outputResults = all;
  91                         return "no";
  92                 }
  93         }
  94         catch(exception& e) {
  95                 errorOut(e, "Slayer", "getResults");
  96                 exit(1);
  97         }
  98 }
  99 /***********************************************************************/
 100 vector<data_struct> Slayer::runBellerophon(Sequence* q, Sequence* pA, Sequence* pB) {
 101         try{
 102
 103                 vector<data_struct> data;
 104 cout << q->getName() << '\t' << q->getAligned().length() << endl;
 105                 //vertical filter
 106                 vector<Sequence*> temp;
 107                 temp.push_back(q); temp.push_back(pA); temp.push_back(pB);
 108                 verticalFilter(temp);
 109
 110                 //get these to avoid numerous function calls
 111                 string query = q->getAligned();
 112                 string parentA = pA->getAligned();
 113                 string parentB = pB->getAligned();
 114                 int length = query.length();
 115 cout << q->getName() << '\t' << length << endl;
 116
 117                 //check window size
 118                 if (length < (2*windowSize+windowStep)) {
 119                         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."); mothurOutEndLine();
 120                         windowSize = length / 4;
 121                 }
 122
 123                 for (int i = windowSize-1; i <= (length - windowSize); i += windowStep) {
 124
 125                         int breakpoint = i;
 126                         int leftLength = breakpoint + 1;
 127                         int rightLength = length - leftLength;
 128
 129                         float QLA = computePercentID(query, parentA, 0, breakpoint);
 130                         float QRB = computePercentID(query, parentB, breakpoint+1, length - 1);
 131
 132                         float QLB = computePercentID(query, parentB, 0, breakpoint);
 133                         float QRA = computePercentID(query, parentA, breakpoint+1, length - 1);
 134
 135                         float LAB = computePercentID(parentA, parentB, 0, breakpoint);
 136                         float RAB = computePercentID(parentA, parentB, breakpoint+1, length - 1);
 137
 138                         float AB = ((LAB*leftLength) + (RAB*rightLength)) / (float) length;
 139                         float QA = ((QLA*leftLength) + (QRA*rightLength)) / (float) length;
 140                         float QB = ((QLB*leftLength) + (QRB*rightLength)) / (float) length;
 141
 142                         float QLA_QRB = ((QLA*leftLength) + (QRB*rightLength)) / (float) length;
 143                         float QLB_QRA = ((QLB*leftLength) + (QRA*rightLength)) / (float) length;
 144
 145                         //in original and not used
 146                         //float avgQA_QB = ((QA*leftLength) + (QB*rightLength)) / (float) length;
 147
 148                         float divR_QLA_QRB = min((QLA_QRB/QA), (QLA_QRB/QB));
 149
 150                         float divR_QLB_QRA = min((QLB_QRA/QA), (QLB_QRA/QB));
 151
 152                         //is one of them above the
 153                         if (divR_QLA_QRB >= divRThreshold || divR_QLB_QRA >= divRThreshold) {
 154
 155                                 if (((QLA_QRB > QA) && (QLA_QRB > QB) && (QLA >= parentFragmentThreshold) && (QRB >= parentFragmentThreshold))  ||
 156                                         ((QLB_QRA > QA) && (QLB_QRA > QB) && (QLB >=parentFragmentThreshold) && (QRA >= parentFragmentThreshold)))  {
 157
 158                                         data_struct member;
 159
 160                                         member.divr_qla_qrb = divR_QLA_QRB;
 161                                         member.divr_qlb_qra = divR_QLB_QRA;
 162                                         member.qla_qrb = QLA_QRB;
 163                                         member.qlb_qra = QLB_QRA;
 164                                         member.qla = QLA;
 165                                         member.qrb = QRB;
 166                                         member.ab = AB;
 167                                         member.qa = QA;
 168                                         member.qb = QB;
 169                                         member.lab = LAB;
 170                                         member.rab = RAB;
 171                                         member.qra = QRA;
 172                                         member.qlb = QLB;
 173                                         member.winLStart = 0;
 174                                         member.winLEnd = breakpoint;
 175                                         member.winRStart = breakpoint+1;
 176                                         member.winREnd = length-1;
 177                                         member.querySeq = *(q);
 178                                         member.parentA = *(pA);
 179                                         member.parentB = *(pB);
 180                                         member.bsa = 0;
 181                                         member.bsb = 0;
 182                                         member.bsMax = 0;
 183                                         member.chimeraMax = 0;
 184
 185                                         data.push_back(member);
 186
 187                                 }//if
 188                         }//if
 189                 }//for
 190
 191                 return data;
 192
 193         }
 194         catch(exception& e) {
 195                 errorOut(e, "Slayer", "runBellerophon");
 196                 exit(1);
 197         }
 198 }
 199 /***********************************************************************/
 200 vector<snps> Slayer::getSNPS(string parentA, string query, string parentB, int left, int right) {
 201         try {
 202
 203                 vector<snps> data;
 204
 205                 for (int i = left; i <= right; i++) {
 206
 207                         char A = parentA[i];
 208                         char Q = query[i];
 209                         char B = parentB[i];
 210
 211                         if ((A != Q) || (B != Q)) {
 212                                 snps member;
 213                                 member.queryChar = Q;
 214                                 member.parentAChar = A;
 215                                 member.parentBChar = B;
 216
 217                                 data.push_back(member);
 218                         }
 219                 }
 220
 221                 return data;
 222
 223         }
 224         catch(exception& e) {
 225                 errorOut(e, "Slayer", "getSNPS");
 226                 exit(1);
 227         }
 228 }
 229 /***********************************************************************/
 230 void Slayer::bootstrapSNPS(vector<snps> left, vector<snps> right, float& BSA, float& BSB) {
 231         try {
 232
 233                 srand((unsigned)time( NULL ));
 234
 235                 int count_A = 0; // sceneario QLA,QRB supported
 236                 int count_B = 0; // sceneario QLB,QRA supported
 237
 238                 int numLeft = max(1, int(left.size()/10 +0.5));
 239                 int numRight = max(1, int(right.size()/10 + 0.5));
 240
 241                 for (int i = 0; i < 100; i++) {
 242                         //random sampling with replacement.
 243
 244                         vector<snps> selectedLeft;
 245                         for (int j = 0; j < numLeft; j++) {
 246                                 int index = int(rand() % left.size());
 247                                 selectedLeft.push_back(left[index]);
 248                         }
 249
 250                         vector<snps> selectedRight;
 251                         for (int j = 0; j < numRight; j++) {
 252                                 int index = int(rand() % right.size());
 253                                 selectedRight.push_back(right[index]);
 254                         }
 255
 256                         /* A  ------------------------------------------
 257                         #       QLA                     QRA
 258                         # Q  ------------------------------------------
 259                         #                      |
 260                         #                      |
 261                         # Q  ------------------------------------------
 262                         #       QLB                     QRB
 263                         # B  ------------------------------------------ */
 264
 265
 266                         float QLA = snpQA(selectedLeft);
 267                         float QRA = snpQA(selectedRight);
 268
 269                         float QLB = snpQB(selectedLeft);
 270                         float QRB = snpQB(selectedRight);
 271
 272                         //in original - not used - not sure why?
 273                         //float ALB = snpAB(selectedLeft);
 274                         //float ARB = snpAB(selectedRight);
 275
 276                         if ((QLA > QLB) && (QRB > QRA)) {
 277                                 count_A++;
 278                         }
 279
 280                         if ((QLB > QLA) && (QRA > QRB)) {
 281                                 count_B++;
 282                         }
 283
 284                 }
 285
 286                 BSA = (float) count_A;
 287                 BSB = (float) count_B;
 288
 289         }
 290         catch(exception& e) {
 291                 errorOut(e, "Slayer", "bootstrapSNPS");
 292                 exit(1);
 293         }
 294 }
 295 /***********************************************************************/
 296 float Slayer::snpQA(vector<snps> data) {
 297         try {
 298
 299                 int numIdentical = 0;
 300
 301                 for (int i = 0; i < data.size(); i++) {
 302                         if (data[i].parentAChar == data[i].queryChar) {
 303                                 numIdentical++;
 304                         }
 305                 }
 306
 307                 float percentID = (numIdentical / data.size()) * 100;
 308
 309                 return percentID;
 310         }
 311         catch(exception& e) {
 312                 errorOut(e, "Slayer", "snpQA");
 313                 exit(1);
 314         }
 315 }
 316 /***********************************************************************/
 317 float Slayer::snpQB(vector<snps> data) {
 318         try {
 319
 320                 int numIdentical = 0;
 321
 322                 for (int i = 0; i < data.size(); i++) {
 323                         if (data[i].parentBChar == data[i].queryChar) {
 324                                 numIdentical++;
 325                         }
 326                 }
 327
 328                 float percentID = (numIdentical / data.size()) * 100;
 329
 330                 return percentID;
 331
 332         }
 333         catch(exception& e) {
 334                 errorOut(e, "Slayer", "snpQB");
 335                 exit(1);
 336         }
 337 }
 338 /***********************************************************************/
 339 float Slayer::snpAB(vector<snps> data) {
 340         try {
 341                 int numIdentical = 0;
 342
 343                 for (int i = 0; i < data.size(); i++) {
 344                         if (data[i].parentAChar == data[i].parentBChar) {
 345                                 numIdentical++;
 346                         }
 347                 }
 348
 349                 float percentID = (numIdentical / data.size()) * 100;
 350
 351                 return percentID;
 352
 353         }
 354         catch(exception& e) {
 355                 errorOut(e, "Slayer", "snpAB");
 356                 exit(1);
 357         }
 358 }
 359 /***********************************************************************/
 360 float Slayer::computePercentID(string queryFrag, string parent, int left, int right) {
 361         try {
 362                 int total = 0;
 363                 int matches = 0;
 364
 365                 for (int i = left; i <= right; i++) {
 366                         total++;
 367                         if (queryFrag[i] == parent[i]) {
 368                                 matches++;
 369                         }
 370                 }
 371
 372                 float percentID =( matches/(float)total) * 100;
 373
 374                 return percentID;
 375         }
 376         catch(exception& e) {
 377                 errorOut(e, "Slayer", "computePercentID");
 378                 exit(1);
 379         }
 380 }
 381 /***********************************************************************/
 382 //remove columns that contain any gaps
 383 void Slayer::verticalFilter(vector<Sequence*> seqs) {
 384         try {
 385                 vector<int> gaps;       gaps.resize(seqs[0]->getAligned().length(), 0);
 386
 387                 string filterString = (string(seqs[0]->getAligned().length(), '1'));
 388
 389                 //for each sequence
 390                 for (int i = 0; i < seqs.size(); i++) {
 391
 392                         string seqAligned = seqs[i]->getAligned();
 393
 394                         for (int j = 0; j < seqAligned.length(); j++) {
 395                                 //if this spot is a gap
 396                                 if ((seqAligned[j] == '-') || (seqAligned[j] == '.') || (toupper(seqAligned[j]) == 'N'))        {       gaps[j]++;      }
 397                         }
 398                 }
 399
 400                 //zero out spot where all sequences have blanks
 401                 int numColRemoved = 0;
 402
 403                 for(int i = 0; i < seqs[0]->getAligned().length(); i++){
 404                         if(gaps[i] != 0)        {       filterString[i] = '0';  numColRemoved++;  }
 405                 }
 406
 407                 //for each sequence
 408                 for (int i = 0; i < seqs.size(); i++) {
 409
 410                         string seqAligned = seqs[i]->getAligned();
 411                         string newAligned = "";
 412
 413                         for (int j = 0; j < seqAligned.length(); j++) {
 414                                 //if this spot is not a gap
 415                                 if (filterString[j] == '1') { newAligned += seqAligned[j]; }
 416                         }
 417
 418                         seqs[i]->setAligned(newAligned);
 419                 }
 420         }
 421         catch(exception& e) {
 422                 errorOut(e, "Slayer", "verticalFilter");
 423                 exit(1);
 424         }
 425 }
 426 /***********************************************************************/