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, string o)  {
  19         try {
  20                 fastafile = name;
  21                 outputDir = o;
  22         }
  23         catch(exception& e) {
  24                 m->errorOut(e, "Bellerophon", "Bellerophon");
  25                 exit(1);
  26         }
  27 }
  28
  29 //***************************************************************************************************************
  30 void Bellerophon::print(ostream& out, ostream& outAcc) {
  31         try {
  32                 int above1 = 0;
  33                 out << "Name\tScore\tLeft\tRight\t" << endl;
  34                 //output prefenence structure to .chimeras file
  35                 for (int i = 0; i < pref.size(); i++) {
  36                         out << pref[i].name << '\t' << setprecision(3) << pref[i].score[0] << '\t' << pref[i].leftParent[0] << '\t' << pref[i].rightParent[0] << endl;
  37
  38                         //calc # of seqs with preference above 1.0
  39                         if (pref[i].score[0] > 1.0) {
  40                                 above1++;
  41                                 outAcc << pref[i].name << endl;
  42                                 m->mothurOut(pref[i].name + " is a suspected chimera at breakpoint " + toString(pref[i].midpoint)); m->mothurOutEndLine();
  43                                 m->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]); m->mothurOutEndLine();
  44                         }
  45                 }
  46
  47                 //output results to screen
  48                 m->mothurOutEndLine();
  49                 m->mothurOut("Sequence with preference score above 1.0: " + toString(above1)); m->mothurOutEndLine();
  50                 int spot;
  51                 spot = pref.size()-1;
  52                 m->mothurOut("Minimum:\t" + toString(pref[spot].score[0])); m->mothurOutEndLine();
  53                 spot = pref.size() * 0.975;
  54                 m->mothurOut("2.5%-tile:\t" + toString(pref[spot].score[0])); m->mothurOutEndLine();
  55                 spot = pref.size() * 0.75;
  56                 m->mothurOut("25%-tile:\t" + toString(pref[spot].score[0])); m->mothurOutEndLine();
  57                 spot = pref.size() * 0.50;
  58                 m->mothurOut("Median: \t" + toString(pref[spot].score[0])); m->mothurOutEndLine();
  59                 spot = pref.size() * 0.25;
  60                 m->mothurOut("75%-tile:\t" + toString(pref[spot].score[0])); m->mothurOutEndLine();
  61                 spot = pref.size() * 0.025;
  62                 m->mothurOut("97.5%-tile:\t" + toString(pref[spot].score[0])); m->mothurOutEndLine();
  63                 spot = 0;
  64                 m->mothurOut("Maximum:\t" + toString(pref[spot].score[0])); m->mothurOutEndLine();
  65
  66         }
  67         catch(exception& e) {
  68                 m->errorOut(e, "Bellerophon", "print");
  69                 exit(1);
  70         }
  71 }
  72
  73 //********************************************************************************************************************
  74 //sorts highest score to lowest
  75 inline bool comparePref(Preference left, Preference right){
  76         return (left.score[0] > right.score[0]);
  77 }
  78
  79 //***************************************************************************************************************
  80 int Bellerophon::getChimeras() {
  81         try {
  82
  83                 //do soft filter
  84                 if (filter)  {
  85                         string optionString = "fasta=" + fastafile + ", soft=50";
  86                         if (outputDir != "") { optionString += ", outputdir=" + outputDir; }
  87
  88                         filterSeqs = new FilterSeqsCommand(optionString);
  89                         filterSeqs->execute();
  90                         delete filterSeqs;
  91
  92                         //reset fastafile to filtered file
  93                         if (outputDir == "") { fastafile = getRootName(fastafile) + "filter.fasta"; }
  94                         else                             { fastafile = outputDir + getRootName(getSimpleName(fastafile)) + "filter.fasta"; }
  95
  96                 }
  97
  98                 distCalculator = new eachGapDist();
  99 cout << "fastafile = " << fastafile << endl;
 100                 //read in sequences
 101                 seqs = readSeqs(fastafile);
 102         cout << "here:"<< endl;
 103                 if (unaligned) { m->mothurOut("Your sequences need to be aligned when you use the bellerophon method."); m->mothurOutEndLine(); return 1;  }
 104
 105                 int numSeqs = seqs.size();
 106
 107                 if (numSeqs == 0) { m->mothurOut("Error in reading you sequences."); m->mothurOutEndLine(); exit(1); }
 108
 109                 //set default window to 25% of sequence length
 110                 string seq0 = seqs[0]->getAligned();
 111                 if (window == 0) { window = seq0.length() / 4;  }
 112                 else if (window > (seq0.length() / 2)) {
 113                         m->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))); m->mothurOutEndLine();
 114                         window = (seq0.length() / 2);
 115                 }
 116
 117                 if (increment > (seqs[0]->getAlignLength() - (2*window))) {
 118                         if (increment != 10) {
 119
 120                                 m->mothurOut("You have selected a increment that is too large. I will use the default."); m->mothurOutEndLine();
 121                                 increment = 10;
 122                                 if (increment > (seqs[0]->getAlignLength() - (2*window))) {  increment = 0;  }
 123
 124                         }else{ increment = 0; }
 125                 }
 126
 127                 if (increment == 0) { iters = 1; }
 128                 else { iters = ((seqs[0]->getAlignLength() - (2*window)) / increment); }
 129
 130                 //initialize pref
 131                 pref.resize(numSeqs);
 132
 133                 for (int i = 0; i < numSeqs; i++ ) {
 134                         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);
 135                         pref[i].name = seqs[i]->getName();
 136                         pref[i].score[0] = 0.0;  pref[i].score[1] = 0.0;
 137                         pref[i].closestLeft[0] = 100000.0;  pref[i].closestLeft[1] = 100000.0;
 138                         pref[i].closestRight[0] = 100000.0;  pref[i].closestRight[1] = 100000.0;
 139                 }
 140
 141                 int midpoint = window;
 142                 int count = 0;
 143                 while (count < iters) {
 144
 145                                 //create 2 vectors of sequences, 1 for left side and one for right side
 146                                 vector<Sequence> left;  vector<Sequence> right;
 147
 148                                 for (int i = 0; i < seqs.size(); i++) {
 149 //cout << "midpoint = " << midpoint << "\twindow = " << window << endl;
 150 //cout << "whole = " << seqs[i]->getAligned().length() << endl;
 151                                         //save left side
 152                                         string seqLeft = seqs[i]->getAligned().substr(midpoint-window, window);
 153                                         Sequence tempLeft;
 154                                         tempLeft.setName(seqs[i]->getName());
 155                                         tempLeft.setAligned(seqLeft);
 156                                         left.push_back(tempLeft);
 157 //cout << "left = " << tempLeft.getAligned().length() << endl;
 158                                         //save right side
 159                                         string seqRight = seqs[i]->getAligned().substr(midpoint, window);
 160                                         Sequence tempRight;
 161                                         tempRight.setName(seqs[i]->getName());
 162                                         tempRight.setAligned(seqRight);
 163                                         right.push_back(tempRight);
 164 //cout << "right = " << seqRight.length() << endl;
 165                                 }
 166
 167                                 //adjust midpoint by increment
 168                                 midpoint += increment;
 169
 170
 171                                 //this should be parallelized
 172                                 //perference = sum of (| distance of my left to sequence j's left - distance of my right to sequence j's right | )
 173                                 //create a matrix containing the distance from left to left and right to right
 174                                 //calculate distances
 175                                 SparseMatrix* SparseLeft = new SparseMatrix();
 176                                 SparseMatrix* SparseRight = new SparseMatrix();
 177
 178                                 createSparseMatrix(0, left.size(), SparseLeft, left);
 179                                 createSparseMatrix(0, right.size(), SparseRight, right);
 180
 181                                 vector<SeqMap> distMapRight;
 182                                 vector<SeqMap> distMapLeft;
 183
 184                                 // Create a data structure to quickly access the distance information.
 185                                 //this is from thallingers reimplementation on get.oturep
 186                                 // It consists of a vector of distance maps, where each map contains
 187                                 // all distances of a certain sequence. Vector and maps are accessed
 188                                 // via the index of a sequence in the distance matrix
 189                                 distMapRight = vector<SeqMap>(numSeqs);
 190                                 distMapLeft = vector<SeqMap>(numSeqs);
 191                                 //cout << "left" << endl << endl;
 192                                 for (MatData currentCell = SparseLeft->begin(); currentCell != SparseLeft->end(); currentCell++) {
 193                                         distMapLeft[currentCell->row][currentCell->column] = currentCell->dist;
 194                                         //cout << " i = " << currentCell->row << " j = " << currentCell->column << " dist = " << currentCell->dist << endl;
 195                                 }
 196                                 //cout << "right" << endl << endl;
 197                                 for (MatData currentCell = SparseRight->begin(); currentCell != SparseRight->end(); currentCell++) {
 198                                         distMapRight[currentCell->row][currentCell->column] = currentCell->dist;
 199                                         //cout << " i = " << currentCell->row << " j = " << currentCell->column << " dist = " << currentCell->dist << endl;
 200                                 }
 201
 202                                 delete SparseLeft;
 203                                 delete SparseRight;
 204
 205                                 //fill preference structure
 206                                 generatePreferences(distMapLeft, distMapRight, midpoint);
 207
 208                                 count++;
 209
 210                 }
 211
 212                 delete distCalculator;
 213
 214                 //rank preference score to eachother
 215                 float dme = 0.0;
 216                 float expectedPercent = 1 / (float) (pref.size());
 217
 218                 for (int i = 0; i < pref.size(); i++) {  dme += pref[i].score[0];  }
 219
 220                 for (int i = 0; i < pref.size(); i++) {
 221
 222                         //gives the actual percentage of the dme this seq adds
 223                         pref[i].score[0] = pref[i].score[0] / dme;
 224
 225                         //how much higher or lower is this than expected
 226                         pref[i].score[0] = pref[i].score[0] / expectedPercent;
 227
 228                 }
 229
 230                 //sort Preferences highest to lowest
 231                 sort(pref.begin(), pref.end(), comparePref);
 232
 233                 return 0;
 234
 235         }
 236         catch(exception& e) {
 237                 m->errorOut(e, "Bellerophon", "getChimeras");
 238                 exit(1);
 239         }
 240 }
 241
 242 /***************************************************************************************************************/
 243 int Bellerophon::createSparseMatrix(int startSeq, int endSeq, SparseMatrix* sparse, vector<Sequence> s){
 244         try {
 245
 246                 for(int i=startSeq; i<endSeq; i++){
 247
 248                         for(int j=0;j<i;j++){
 249
 250                                 distCalculator->calcDist(s[i], s[j]);
 251                                 float dist = distCalculator->getDist();
 252
 253                                 PCell temp(i, j, dist);
 254                                 sparse->addCell(temp);
 255
 256                         }
 257                 }
 258
 259                 return 1;
 260         }
 261         catch(exception& e) {
 262                 m->errorOut(e, "Bellerophon", "createSparseMatrix");
 263                 exit(1);
 264         }
 265 }
 266 /***************************************************************************************************************/
 267 void Bellerophon::generatePreferences(vector<SeqMap> left, vector<SeqMap> right, int mid){
 268         try {
 269
 270                 float dme = 0.0;
 271                 SeqMap::iterator itR;
 272                 SeqMap::iterator itL;
 273
 274                 //initialize pref[i]
 275                 for (int i = 0; i < pref.size(); i++) {
 276                         pref[i].score[1] = 0.0;
 277                         pref[i].closestLeft[1] = 100000.0;
 278                         pref[i].closestRight[1] = 100000.0;
 279                         pref[i].leftParent[1] = "";
 280                         pref[i].rightParent[1] = "";
 281                 }
 282
 283                 for (int i = 0; i < left.size(); i++) {
 284
 285                         SeqMap currentLeft = left[i];    //example i = 3;   currentLeft is a map of 0 to the distance of sequence 3 to sequence 0,
 286                                                                                                 //                                                                              1 to the distance of sequence 3 to sequence 1,
 287                                                                                                 //                                                                              2 to the distance of sequence 3 to sequence 2.
 288                         SeqMap currentRight = right[i];         // same as left but with distances on the right side.
 289
 290                         for (int j = 0; j < i; j++) {
 291
 292                                 itL = currentLeft.find(j);
 293                                 itR = currentRight.find(j);
 294 //cout << " i = " << i << " j = " << j << " distLeft = " << itL->second << endl;
 295 //cout << " i = " << i << " j = " << j << " distright = " << itR->second << endl;
 296
 297                                 //if you can find this entry update the preferences
 298                                 if ((itL != currentLeft.end()) && (itR != currentRight.end())) {
 299
 300                                         if (!correction) {
 301                                                 pref[i].score[1] += abs((itL->second - itR->second));
 302                                                 pref[j].score[1] += abs((itL->second - itR->second));
 303 //cout << "left " << i << " " << j << " = " << itL->second << " right " << i << " " << j << " = " << itR->second << endl;
 304 //cout << "abs = " << abs((itL->second - itR->second)) << endl;
 305 //cout << i << " score = " << pref[i].score[1] << endl;
 306 //cout << j << " score = " << pref[j].score[1] << endl;
 307                                         }else {
 308                                                 pref[i].score[1] += abs((sqrt(itL->second) - sqrt(itR->second)));
 309                                                 pref[j].score[1] += abs((sqrt(itL->second) - sqrt(itR->second)));
 310 //cout << "left " << i << " " << j << " = " << itL->second << " right " << i << " " << j << " = " << itR->second << endl;
 311 //cout << "abs = " << abs((sqrt(itL->second) - sqrt(itR->second))) << endl;
 312 //cout << i << " score = " << pref[i].score[1] << endl;
 313 //cout << j << " score = " << pref[j].score[1] << endl;
 314                                         }
 315 //cout << "pref[" << i << "].closestLeft[1] = " <<      pref[i].closestLeft[1] << " parent = " << pref[i].leftParent[1] << endl;
 316                                         //are you the closest left sequence
 317                                         if (itL->second < pref[i].closestLeft[1]) {
 318
 319                                                 pref[i].closestLeft[1] = itL->second;
 320                                                 pref[i].leftParent[1] = seqs[j]->getName();
 321 //cout << "updating closest left to " << pref[i].leftParent[1] << endl;
 322                                         }
 323 //cout << "pref[" << j << "].closestLeft[1] = " <<      pref[j].closestLeft[1] << " parent = " << pref[j].leftParent[1] << endl;
 324                                         if (itL->second < pref[j].closestLeft[1]) {
 325                                                 pref[j].closestLeft[1] = itL->second;
 326                                                 pref[j].leftParent[1] = seqs[i]->getName();
 327 //cout << "updating closest left to " << pref[j].leftParent[1] << endl;
 328                                         }
 329
 330                                         //are you the closest right sequence
 331                                         if (itR->second < pref[i].closestRight[1]) {
 332                                                 pref[i].closestRight[1] = itR->second;
 333                                                 pref[i].rightParent[1] = seqs[j]->getName();
 334                                         }
 335                                         if (itR->second < pref[j].closestRight[1]) {
 336                                                 pref[j].closestRight[1] = itR->second;
 337                                                 pref[j].rightParent[1] = seqs[i]->getName();
 338                                         }
 339
 340                                 }
 341                         }
 342
 343                 }
 344
 345
 346
 347                 //calculate the dme
 348                 int count0 = 0;
 349                 for (int i = 0; i < pref.size(); i++) {  dme += pref[i].score[1];  if (pref[i].score[1] == 0.0) { count0++; }  }
 350
 351                 float expectedPercent = 1 / (float) (pref.size() - count0);
 352 //cout << endl << "dme = " << dme << endl;
 353                 //recalculate prefernences based on dme
 354                 for (int i = 0; i < pref.size(); i++) {
 355 //cout << "unadjusted pref " << i << " = " << pref[i].score[1] << endl;
 356                         // gives the actual percentage of the dme this seq adds
 357                         pref[i].score[1] = pref[i].score[1] / dme;
 358
 359                         //how much higher or lower is this than expected
 360                         pref[i].score[1] = pref[i].score[1] / expectedPercent;
 361
 362                         //pref[i].score[1] = dme / (dme - 2 * pref[i].score[1]);
 363
 364                         //so a non chimeric sequence would be around 1, and a chimeric would be signifigantly higher.
 365 //cout << "adjusted pref " << i << " = " << pref[i].score[1] << endl;
 366                 }
 367
 368                 //is this score bigger then the last score
 369                 for (int i = 0; i < pref.size(); i++) {
 370
 371                         //update biggest score
 372                         if (pref[i].score[1] > pref[i].score[0]) {
 373                                 pref[i].score[0] = pref[i].score[1];
 374                                 pref[i].leftParent[0] = pref[i].leftParent[1];
 375                                 pref[i].rightParent[0] = pref[i].rightParent[1];
 376                                 pref[i].closestLeft[0] = pref[i].closestLeft[1];
 377                                 pref[i].closestRight[0] = pref[i].closestRight[1];
 378                                 pref[i].midpoint = mid;
 379                         }
 380
 381                 }
 382
 383         }
 384         catch(exception& e) {
 385                 m->errorOut(e, "Bellerophon", "generatePreferences");
 386                 exit(1);
 387         }
 388 }
 389 /**************************************************************************************************/
 390