bellerophon.cpp

   1 /*
   2  *  bellerophon.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 7/9/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "bellerophon.h"
  11 #include "eachgapdist.h"
  12 #include "ignoregaps.h"
  13 #include "onegapdist.h"
  14
  15
  16 //***************************************************************************************************************
  17
  18 Bellerophon::Bellerophon(string name) {
  19         try {
  20                 fastafile = name;
  21         }
  22         catch(exception& e) {
  23                 errorOut(e, "Bellerophon", "Bellerophon");
  24                 exit(1);
  25         }
  26 }
  27
  28 //***************************************************************************************************************
  29 void Bellerophon::print(ostream& out) {
  30         try {
  31                 int above1 = 0;
  32                 out << "Name\tScore\tLeft\tRight\t" << endl;
  33                 //output prefenence structure to .chimeras file
  34                 for (int i = 0; i < pref.size(); i++) {
  35                         out << pref[i].name << '\t' << setprecision(3) << pref[i].score[0] << '\t' << pref[i].leftParent[0] << '\t' << pref[i].rightParent[0] << endl;
  36
  37                         //calc # of seqs with preference above 1.0
  38                         if (pref[i].score[0] > 1.0) {
  39                                 above1++;
  40                                 mothurOut(pref[i].name + " is a suspected chimera at breakpoint " + toString(pref[i].midpoint)); mothurOutEndLine();
  41                                 mothurOut("It's score is " + toString(pref[i].score[0]) + " with suspected left parent " + pref[i].leftParent[0] + " and right parent " + pref[i].rightParent[0]); mothurOutEndLine();
  42                         }
  43                 }
  44
  45                 //output results to screen
  46                 mothurOutEndLine();
  47                 mothurOut("Sequence with preference score above 1.0: " + toString(above1)); mothurOutEndLine();
  48                 int spot;
  49                 spot = pref.size()-1;
  50                 mothurOut("Minimum:\t" + toString(pref[spot].score[0])); mothurOutEndLine();
  51                 spot = pref.size() * 0.975;
  52                 mothurOut("2.5%-tile:\t" + toString(pref[spot].score[0])); mothurOutEndLine();
  53                 spot = pref.size() * 0.75;
  54                 mothurOut("25%-tile:\t" + toString(pref[spot].score[0])); mothurOutEndLine();
  55                 spot = pref.size() * 0.50;
  56                 mothurOut("Median: \t" + toString(pref[spot].score[0])); mothurOutEndLine();
  57                 spot = pref.size() * 0.25;
  58                 mothurOut("75%-tile:\t" + toString(pref[spot].score[0])); mothurOutEndLine();
  59                 spot = pref.size() * 0.025;
  60                 mothurOut("97.5%-tile:\t" + toString(pref[spot].score[0])); mothurOutEndLine();
  61                 spot = 0;
  62                 mothurOut("Maximum:\t" + toString(pref[spot].score[0])); mothurOutEndLine();
  63
  64         }
  65         catch(exception& e) {
  66                 errorOut(e, "Bellerophon", "print");
  67                 exit(1);
  68         }
  69 }
  70
  71 //********************************************************************************************************************
  72 //sorts highest score to lowest
  73 inline bool comparePref(Preference left, Preference right){
  74         return (left.score[0] > right.score[0]);
  75 }
  76
  77 //***************************************************************************************************************
  78 int Bellerophon::getChimeras() {
  79         try {
  80
  81                 //do soft filter
  82                 if (filter)  {
  83                         string optionString = "fasta=" + fastafile + ", soft=50";
  84                         filterSeqs = new FilterSeqsCommand(optionString);
  85                         filterSeqs->execute();
  86                         delete filterSeqs;
  87
  88                         //reset fastafile to filtered file
  89                         fastafile = getRootName(fastafile) + "filter.fasta";
  90                 }
  91
  92                 distCalculator = new eachGapDist();
  93
  94                 //read in sequences
  95                 seqs = readSeqs(fastafile);
  96
  97                 if (unaligned) { mothurOut("Your sequences need to be aligned when you use the bellerophon method."); mothurOutEndLine(); return 1;  }
  98
  99                 int numSeqs = seqs.size();
 100
 101                 if (numSeqs == 0) { mothurOut("Error in reading you sequences."); mothurOutEndLine(); exit(1); }
 102
 103                 //set default window to 25% of sequence length
 104                 string seq0 = seqs[0]->getAligned();
 105                 if (window == 0) { window = seq0.length() / 4;  }
 106                 else if (window > (seq0.length() / 2)) {
 107                         mothurOut("Your sequence length is = " + toString(seq0.length()) + ". You have selected a window size greater than the length of half your aligned sequence. I will run it with a window size of " + toString((seq0.length() / 2))); mothurOutEndLine();
 108                         window = (seq0.length() / 2);
 109                 }
 110
 111                 if (increment > (seqs[0]->getAlignLength() - (2*window))) {
 112                         if (increment != 10) {
 113
 114                                 mothurOut("You have selected a increment that is too large. I will use the default."); mothurOutEndLine();
 115                                 increment = 10;
 116                                 if (increment > (seqs[0]->getAlignLength() - (2*window))) {  increment = 0;  }
 117
 118                         }else{ increment = 0; }
 119                 }
 120
 121                 if (increment == 0) { iters = 1; }
 122                 else { iters = ((seqs[0]->getAlignLength() - (2*window)) / increment); }
 123
 124                 //initialize pref
 125                 pref.resize(numSeqs);
 126
 127                 for (int i = 0; i < numSeqs; i++ ) {
 128                         pref[i].leftParent.resize(2); pref[i].rightParent.resize(2); pref[i].score.resize(2);   pref[i].closestLeft.resize(2); pref[i].closestRight.resize(3);
 129                         pref[i].name = seqs[i]->getName();
 130                         pref[i].score[0] = 0.0;  pref[i].score[1] = 0.0;
 131                         pref[i].closestLeft[0] = 100000.0;  pref[i].closestLeft[1] = 100000.0;
 132                         pref[i].closestRight[0] = 100000.0;  pref[i].closestRight[1] = 100000.0;
 133                 }
 134
 135                 int midpoint = window;
 136                 int count = 0;
 137                 while (count < iters) {
 138
 139                                 //create 2 vectors of sequences, 1 for left side and one for right side
 140                                 vector<Sequence> left;  vector<Sequence> right;
 141
 142                                 for (int i = 0; i < seqs.size(); i++) {
 143 //cout << "midpoint = " << midpoint << "\twindow = " << window << endl;
 144 //cout << "whole = " << seqs[i]->getAligned().length() << endl;
 145                                         //save left side
 146                                         string seqLeft = seqs[i]->getAligned().substr(midpoint-window, window);
 147                                         Sequence tempLeft;
 148                                         tempLeft.setName(seqs[i]->getName());
 149                                         tempLeft.setAligned(seqLeft);
 150                                         left.push_back(tempLeft);
 151 //cout << "left = " << tempLeft.getAligned().length() << endl;
 152                                         //save right side
 153                                         string seqRight = seqs[i]->getAligned().substr(midpoint, window);
 154                                         Sequence tempRight;
 155                                         tempRight.setName(seqs[i]->getName());
 156                                         tempRight.setAligned(seqRight);
 157                                         right.push_back(tempRight);
 158 //cout << "right = " << seqRight.length() << endl;
 159                                 }
 160
 161                                 //adjust midpoint by increment
 162                                 midpoint += increment;
 163
 164
 165                                 //this should be parallelized
 166                                 //perference = sum of (| distance of my left to sequence j's left - distance of my right to sequence j's right | )
 167                                 //create a matrix containing the distance from left to left and right to right
 168                                 //calculate distances
 169                                 SparseMatrix* SparseLeft = new SparseMatrix();
 170                                 SparseMatrix* SparseRight = new SparseMatrix();
 171
 172                                 createSparseMatrix(0, left.size(), SparseLeft, left);
 173                                 createSparseMatrix(0, right.size(), SparseRight, right);
 174
 175                                 vector<SeqMap> distMapRight;
 176                                 vector<SeqMap> distMapLeft;
 177
 178                                 // Create a data structure to quickly access the distance information.
 179                                 //this is from thallingers reimplementation on get.oturep
 180                                 // It consists of a vector of distance maps, where each map contains
 181                                 // all distances of a certain sequence. Vector and maps are accessed
 182                                 // via the index of a sequence in the distance matrix
 183                                 distMapRight = vector<SeqMap>(numSeqs);
 184                                 distMapLeft = vector<SeqMap>(numSeqs);
 185                                 //cout << "left" << endl << endl;
 186                                 for (MatData currentCell = SparseLeft->begin(); currentCell != SparseLeft->end(); currentCell++) {
 187                                         distMapLeft[currentCell->row][currentCell->column] = currentCell->dist;
 188                                         //cout << " i = " << currentCell->row << " j = " << currentCell->column << " dist = " << currentCell->dist << endl;
 189                                 }
 190                                 //cout << "right" << endl << endl;
 191                                 for (MatData currentCell = SparseRight->begin(); currentCell != SparseRight->end(); currentCell++) {
 192                                         distMapRight[currentCell->row][currentCell->column] = currentCell->dist;
 193                                         //cout << " i = " << currentCell->row << " j = " << currentCell->column << " dist = " << currentCell->dist << endl;
 194                                 }
 195
 196                                 delete SparseLeft;
 197                                 delete SparseRight;
 198
 199                                 //fill preference structure
 200                                 generatePreferences(distMapLeft, distMapRight, midpoint);
 201
 202                                 count++;
 203
 204                 }
 205
 206                 delete distCalculator;
 207
 208                 //rank preference score to eachother
 209                 float dme = 0.0;
 210                 float expectedPercent = 1 / (float) (pref.size());
 211
 212                 for (int i = 0; i < pref.size(); i++) {  dme += pref[i].score[0];  }
 213
 214                 for (int i = 0; i < pref.size(); i++) {
 215
 216                         //gives the actual percentage of the dme this seq adds
 217                         pref[i].score[0] = pref[i].score[0] / dme;
 218
 219                         //how much higher or lower is this than expected
 220                         pref[i].score[0] = pref[i].score[0] / expectedPercent;
 221
 222                 }
 223
 224                 //sort Preferences highest to lowest
 225                 sort(pref.begin(), pref.end(), comparePref);
 226
 227                 return 0;
 228
 229         }
 230         catch(exception& e) {
 231                 errorOut(e, "Bellerophon", "getChimeras");
 232                 exit(1);
 233         }
 234 }
 235
 236 /***************************************************************************************************************/
 237 int Bellerophon::createSparseMatrix(int startSeq, int endSeq, SparseMatrix* sparse, vector<Sequence> s){
 238         try {
 239
 240                 for(int i=startSeq; i<endSeq; i++){
 241
 242                         for(int j=0;j<i;j++){
 243
 244                                 distCalculator->calcDist(s[i], s[j]);
 245                                 float dist = distCalculator->getDist();
 246
 247                                 PCell temp(i, j, dist);
 248                                 sparse->addCell(temp);
 249
 250                         }
 251                 }
 252
 253                 return 1;
 254         }
 255         catch(exception& e) {
 256                 errorOut(e, "Bellerophon", "createSparseMatrix");
 257                 exit(1);
 258         }
 259 }
 260 /***************************************************************************************************************/
 261 void Bellerophon::generatePreferences(vector<SeqMap> left, vector<SeqMap> right, int mid){
 262         try {
 263
 264                 float dme = 0.0;
 265                 SeqMap::iterator itR;
 266                 SeqMap::iterator itL;
 267
 268                 //initialize pref[i]
 269                 for (int i = 0; i < pref.size(); i++) {
 270                         pref[i].score[1] = 0.0;
 271                         pref[i].closestLeft[1] = 100000.0;
 272                         pref[i].closestRight[1] = 100000.0;
 273                         pref[i].leftParent[1] = "";
 274                         pref[i].rightParent[1] = "";
 275                 }
 276
 277                 for (int i = 0; i < left.size(); i++) {
 278
 279                         SeqMap currentLeft = left[i];    //example i = 3;   currentLeft is a map of 0 to the distance of sequence 3 to sequence 0,
 280                                                                                                 //                                                                              1 to the distance of sequence 3 to sequence 1,
 281                                                                                                 //                                                                              2 to the distance of sequence 3 to sequence 2.
 282                         SeqMap currentRight = right[i];         // same as left but with distances on the right side.
 283
 284                         for (int j = 0; j < i; j++) {
 285
 286                                 itL = currentLeft.find(j);
 287                                 itR = currentRight.find(j);
 288 //cout << " i = " << i << " j = " << j << " distLeft = " << itL->second << endl;
 289 //cout << " i = " << i << " j = " << j << " distright = " << itR->second << endl;
 290
 291                                 //if you can find this entry update the preferences
 292                                 if ((itL != currentLeft.end()) && (itR != currentRight.end())) {
 293
 294                                         if (!correction) {
 295                                                 pref[i].score[1] += abs((itL->second - itR->second));
 296                                                 pref[j].score[1] += abs((itL->second - itR->second));
 297 //cout << "left " << i << " " << j << " = " << itL->second << " right " << i << " " << j << " = " << itR->second << endl;
 298 //cout << "abs = " << abs((itL->second - itR->second)) << endl;
 299 //cout << i << " score = " << pref[i].score[1] << endl;
 300 //cout << j << " score = " << pref[j].score[1] << endl;
 301                                         }else {
 302                                                 pref[i].score[1] += abs((sqrt(itL->second) - sqrt(itR->second)));
 303                                                 pref[j].score[1] += abs((sqrt(itL->second) - sqrt(itR->second)));
 304 //cout << "left " << i << " " << j << " = " << itL->second << " right " << i << " " << j << " = " << itR->second << endl;
 305 //cout << "abs = " << abs((sqrt(itL->second) - sqrt(itR->second))) << endl;
 306 //cout << i << " score = " << pref[i].score[1] << endl;
 307 //cout << j << " score = " << pref[j].score[1] << endl;
 308                                         }
 309 //cout << "pref[" << i << "].closestLeft[1] = " <<      pref[i].closestLeft[1] << " parent = " << pref[i].leftParent[1] << endl;
 310                                         //are you the closest left sequence
 311                                         if (itL->second < pref[i].closestLeft[1]) {
 312
 313                                                 pref[i].closestLeft[1] = itL->second;
 314                                                 pref[i].leftParent[1] = seqs[j]->getName();
 315 //cout << "updating closest left to " << pref[i].leftParent[1] << endl;
 316                                         }
 317 //cout << "pref[" << j << "].closestLeft[1] = " <<      pref[j].closestLeft[1] << " parent = " << pref[j].leftParent[1] << endl;
 318                                         if (itL->second < pref[j].closestLeft[1]) {
 319                                                 pref[j].closestLeft[1] = itL->second;
 320                                                 pref[j].leftParent[1] = seqs[i]->getName();
 321 //cout << "updating closest left to " << pref[j].leftParent[1] << endl;
 322                                         }
 323
 324                                         //are you the closest right sequence
 325                                         if (itR->second < pref[i].closestRight[1]) {
 326                                                 pref[i].closestRight[1] = itR->second;
 327                                                 pref[i].rightParent[1] = seqs[j]->getName();
 328                                         }
 329                                         if (itR->second < pref[j].closestRight[1]) {
 330                                                 pref[j].closestRight[1] = itR->second;
 331                                                 pref[j].rightParent[1] = seqs[i]->getName();
 332                                         }
 333
 334                                 }
 335                         }
 336
 337                 }
 338
 339
 340
 341                 //calculate the dme
 342                 int count0 = 0;
 343                 for (int i = 0; i < pref.size(); i++) {  dme += pref[i].score[1];  if (pref[i].score[1] == 0.0) { count0++; }  }
 344
 345                 float expectedPercent = 1 / (float) (pref.size() - count0);
 346 //cout << endl << "dme = " << dme << endl;
 347                 //recalculate prefernences based on dme
 348                 for (int i = 0; i < pref.size(); i++) {
 349 //cout << "unadjusted pref " << i << " = " << pref[i].score[1] << endl;
 350                         // gives the actual percentage of the dme this seq adds
 351                         pref[i].score[1] = pref[i].score[1] / dme;
 352
 353                         //how much higher or lower is this than expected
 354                         pref[i].score[1] = pref[i].score[1] / expectedPercent;
 355
 356                         //pref[i].score[1] = dme / (dme - 2 * pref[i].score[1]);
 357
 358                         //so a non chimeric sequence would be around 1, and a chimeric would be signifigantly higher.
 359 //cout << "adjusted pref " << i << " = " << pref[i].score[1] << endl;
 360                 }
 361
 362                 //is this score bigger then the last score
 363                 for (int i = 0; i < pref.size(); i++) {
 364
 365                         //update biggest score
 366                         if (pref[i].score[1] > pref[i].score[0]) {
 367                                 pref[i].score[0] = pref[i].score[1];
 368                                 pref[i].leftParent[0] = pref[i].leftParent[1];
 369                                 pref[i].rightParent[0] = pref[i].rightParent[1];
 370                                 pref[i].closestLeft[0] = pref[i].closestLeft[1];
 371                                 pref[i].closestRight[0] = pref[i].closestRight[1];
 372                                 pref[i].midpoint = mid;
 373                         }
 374
 375                 }
 376
 377         }
 378         catch(exception& e) {
 379                 errorOut(e, "Bellerophon", "generatePreferences");
 380                 exit(1);
 381         }
 382 }
 383 /**************************************************************************************************/
 384