onegapignore.h

   1 #ifndef ONEIGNOREGAPS_H
   2 #define ONEIGNOREGAPS_H
   3 /*
   4  *  onegapignore.h
   5  *  Mothur
   6  *
   7  *  Created by Sarah Westcott on 5/7/09.
   8  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   9  *
  10  */
  11
  12
  13 #include "dist.h"
  14
  15 /**************************************************************************************************/
  16
  17 class oneGapIgnoreTermGapDist : public Dist {
  18
  19 public:
  20
  21         oneGapIgnoreTermGapDist() {}
  22
  23         void calcDist(Sequence A, Sequence B){
  24
  25                 int difference = 0;
  26                 int openGapA = 0;
  27                 int openGapB = 0;
  28                 int minLength = 0;
  29                 int start = 0;
  30                 int end = 0;
  31                 bool overlap = false;
  32
  33                 string seqA = A.getAligned();
  34                 string seqB = B.getAligned();
  35                 int alignLength = seqA.length();
  36
  37                 // this assumes that sequences start and end with '.'s instead of'-'s.
  38                 for(int i=0;i<alignLength;i++){
  39                         if(seqA[i] != '.' && seqB[i] != '.' && seqA[i] != '-' && seqB[i] != '-' ){
  40                                 start = i;
  41 //                              cout << "start: " << start << endl;
  42                                 overlap = true;
  43                                 break;
  44                         }
  45                 }
  46                 for(int i=alignLength-1;i>=0;i--){
  47                         if(seqA[i] != '.' && seqB[i] != '.' && seqA[i] != '-' && seqB[i] != '-' ){
  48                                 end = i;
  49 //                              cout << "end: " << end << endl;
  50                                 overlap = true;
  51                                 break;
  52                         }
  53                 }
  54
  55                 for(int i=start;i<=end;i++){
  56                         if(seqA[i] == '-' && seqB[i] == '-'){   ;       }
  57                         else if(seqB[i] != '-' && seqA[i] == '-'){
  58                                 if(openGapA == 0){
  59                                         difference++;
  60                                         minLength++;
  61                                         openGapA = 1;
  62                                         openGapB = 0;
  63                                 }
  64                         }
  65                         else if(seqA[i] != '-' && seqB[i] == '-'){
  66                                 if(openGapB == 0){
  67                                         difference++;
  68                                         minLength++;
  69                                         openGapA = 0;
  70                                         openGapB = 1;
  71                                 }
  72                         }
  73                         else if(seqA[i] != '-' && seqB[i] != '-'){
  74                                 if(seqA[i] != seqB[i]){
  75                                         difference++;
  76                                 }
  77                                 minLength++;
  78                                 openGapA = 0;
  79                                 openGapB = 0;
  80                         }
  81                 }
  82
  83                 //non-overlapping sequences
  84                 if (!overlap) { minLength = 0; }
  85
  86                 if(minLength == 0)      {       dist = 1.0000;                                                  }
  87                 else                            {       dist = (double)difference / minLength;  }
  88         }
  89
  90 };
  91
  92 /**************************************************************************************************/
  93
  94 #endif
  95