needlemanoverlap.cpp

   1 /*
   2  *  needleman.cpp
   3  *
   4  *
   5  *  Created by Pat Schloss on 12/15/08.
   6  *  Copyright 2008 Patrick D. Schloss. All rights reserved.
   7  *
   8  *      This class is an Alignment child class that implements the Gotoh pairwise alignment algorithm as described in:
   9  *
  10  *              Gotoh O. 1982.  An improved algorithm for matching biological sequences.  J. Mol. Biol.  162:705-8.
  11  *              Myers, EW & Miller, W.  1988.  Optimal alignments in linear space.  Comput Appl Biosci. 4:11-7.
  12  *
  13  *      This method is nice because it allows for an affine gap penalty to be assessed, which is analogous to what is used
  14  *      in blast and is an alternative to Needleman-Wunsch, which only charges the same penalty for each gap position.
  15  *      Because this method typically has problems at the ends when two sequences do not full overlap, we employ a separate
  16  *      method to fix the ends (see Overlap class documentation)
  17  *
  18  */
  19
  20 #include "alignmentcell.hpp"
  21 #include "alignment.hpp"
  22 #include "overlap.hpp"
  23 #include "needlemanoverlap.hpp"
  24
  25 /**************************************************************************************************/
  26
  27 NeedlemanOverlap::NeedlemanOverlap(float gO, float m, float mm, int r) ://      note that we don't have a gap extend
  28 gap(gO), match(m), mismatch(mm), Alignment(r) {                                                 //      the gap openning penalty is assessed for
  29         try {                                                                                                                                   //      every gapped position
  30                 for(int i=1;i<nCols;i++){
  31                         alignment[0][i].prevCell = 'l';                                 //      initialize first row by pointing all poiters to the left
  32                         alignment[0][i].cValue = 0;                                             //      and the score to zero
  33                 }
  34
  35                 for(int i=1;i<nRows;i++){
  36                         alignment[i][0].prevCell = 'u';                                 //      initialize first column by pointing all poiters upwards
  37                         alignment[i][0].cValue = 0;                                             //      and the score to zero
  38                 }
  39
  40         }
  41         catch(exception& e) {
  42                 errorOut(e, "NeedlemanOverlap", "NeedlemanOverlap");
  43                 exit(1);
  44         }
  45
  46
  47 }
  48 /**************************************************************************************************/
  49
  50 NeedlemanOverlap::~NeedlemanOverlap(){  /*      do nothing      */      }
  51
  52 /**************************************************************************************************/
  53
  54 void NeedlemanOverlap::align(string A, string B){
  55         try {
  56                 seqA = ' ' + A; lA = seqA.length();             //      algorithm requires a dummy space at the beginning of each string
  57                 seqB = ' ' + B; lB = seqB.length();             //      algorithm requires a dummy space at the beginning of each string
  58
  59                 if (lA > nRows) { mothurOut("Your one of your candidate sequences is longer than you longest template sequence."); mothurOutEndLine();  }
  60
  61                 for(int i=1;i<lB;i++){                                  //      This code was largely translated from Perl code provided in Ex 3.1
  62                         for(int j=1;j<lA;j++){                          //      of the O'Reilly BLAST book.  I found that the example output had a
  63                                 //      number of errors
  64                                 float diagonal;
  65                                 if(seqB[i] == seqA[j])  {       diagonal = alignment[i-1][j-1].cValue + match;          }
  66                                 else                                    {       diagonal = alignment[i-1][j-1].cValue + mismatch;       }
  67
  68                                 float up        = alignment[i-1][j].cValue + gap;
  69                                 float left      = alignment[i][j-1].cValue + gap;
  70
  71                                 if(diagonal >= up){
  72                                         if(diagonal >= left){
  73                                                 alignment[i][j].cValue = diagonal;
  74                                                 alignment[i][j].prevCell = 'd';
  75                                         }
  76                                         else{
  77                                                 alignment[i][j].cValue = left;
  78                                                 alignment[i][j].prevCell = 'l';
  79                                         }
  80                                 }
  81                                 else{
  82                                         if(up >= left){
  83                                                 alignment[i][j].cValue = up;
  84                                                 alignment[i][j].prevCell = 'u';
  85                                         }
  86                                         else{
  87                                                 alignment[i][j].cValue = left;
  88                                                 alignment[i][j].prevCell = 'l';
  89                                         }
  90                                 }
  91                         }
  92                 }
  93                 Overlap over;
  94                 over.setOverlap(alignment, lA, lB, 0);          //      Fix gaps at the beginning and end of the sequences
  95                 traceBack();                                                            //      Traceback the alignment to populate seqAaln and seqBaln
  96
  97         }
  98         catch(exception& e) {
  99                 errorOut(e, "NeedlemanOverlap", "align");
 100                 exit(1);
 101         }
 102
 103 }
 104
 105 /**************************************************************************************************/
 106