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         void calcDist(Sequence A, Sequence B){
  21
  22                 int difference = 0;
  23                 int openGapA = 0;
  24                 int openGapB = 0;
  25                 int minLength = 0;
  26                 int start = 0;
  27                 int end = 0;
  28                 bool overlap = false;
  29
  30                 string seqA = A.getAligned();
  31                 string seqB = B.getAligned();
  32                 int alignLength = seqA.length();
  33
  34                 // this assumes that sequences start and end with '.'s instead of'-'s.
  35                 for(int i=0;i<alignLength;i++){
  36                         if(seqA[i] != '.' && seqB[i] != '.' && seqA[i] != '-' && seqB[i] != '-' ){
  37                                 start = i;
  38 //                              cout << "start: " << start << endl;
  39                                 overlap = true;
  40                                 break;
  41                         }
  42                 }
  43                 for(int i=alignLength-1;i>=0;i--){
  44                         if(seqA[i] != '.' && seqB[i] != '.' && seqA[i] != '-' && seqB[i] != '-' ){
  45                                 end = i;
  46 //                              cout << "end: " << end << endl;
  47                                 overlap = true;
  48                                 break;
  49                         }
  50                 }
  51
  52                 for(int i=start;i<=end;i++){
  53                         if(seqA[i] == '-' && seqB[i] == '-'){   ;       }
  54                         else if(seqB[i] != '-' && seqA[i] == '-'){
  55                                 if(openGapA == 0){
  56                                         difference++;
  57                                         minLength++;
  58                                         openGapA = 1;
  59                                         openGapB = 0;
  60                                 }
  61                         }
  62                         else if(seqA[i] != '-' && seqB[i] == '-'){
  63                                 if(openGapB == 0){
  64                                         difference++;
  65                                         minLength++;
  66                                         openGapA = 0;
  67                                         openGapB = 1;
  68                                 }
  69                         }
  70                         else if(seqA[i] != '-' && seqB[i] != '-'){
  71                                 if(seqA[i] != seqB[i]){
  72                                         difference++;
  73                                 }
  74                                 minLength++;
  75                                 openGapA = 0;
  76                                 openGapB = 0;
  77                         }
  78                 }
  79
  80                 //non-overlapping sequences
  81                 if (!overlap) { minLength = 0; }
  82
  83                 if(minLength == 0)      {       dist = 1.0000;                                                  }
  84                 else                            {       dist = (double)difference / minLength;  }
  85         }
  86
  87 };
  88
  89 /**************************************************************************************************/
  90
  91 #endif
  92