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         oneGapIgnoreTermGapDist(const oneGapIgnoreTermGapDist& ddb) {}
  23
  24         void calcDist(Sequence A, Sequence B){
  25
  26                 int difference = 0;
  27                 int openGapA = 0;
  28                 int openGapB = 0;
  29                 int minLength = 0;
  30                 int start = 0;
  31                 int end = 0;
  32                 bool overlap = false;
  33
  34                 string seqA = A.getAligned();
  35                 string seqB = B.getAligned();
  36                 int alignLength = seqA.length();
  37
  38                 // this assumes that sequences start and end with '.'s instead of'-'s.
  39                 for(int i=0;i<alignLength;i++){
  40                         if(seqA[i] != '.' && seqB[i] != '.' && seqA[i] != '-' && seqB[i] != '-' ){
  41                                 start = i;
  42 //                              cout << "start: " << start << endl;
  43                                 overlap = true;
  44                                 break;
  45                         }
  46                 }
  47                 for(int i=alignLength-1;i>=0;i--){
  48                         if(seqA[i] != '.' && seqB[i] != '.' && seqA[i] != '-' && seqB[i] != '-' ){
  49                                 end = i;
  50 //                              cout << "end: " << end << endl;
  51                                 overlap = true;
  52                                 break;
  53                         }
  54                 }
  55
  56                 for(int i=start;i<=end;i++){
  57                         if(seqA[i] == '-' && seqB[i] == '-'){   ;       }
  58                         else if(seqB[i] != '-' && seqA[i] == '-'){
  59                                 if(openGapA == 0){
  60                                         difference++;
  61                                         minLength++;
  62                                         openGapA = 1;
  63                                         openGapB = 0;
  64                                 }
  65                         }
  66                         else if(seqA[i] != '-' && seqB[i] == '-'){
  67                                 if(openGapB == 0){
  68                                         difference++;
  69                                         minLength++;
  70                                         openGapA = 0;
  71                                         openGapB = 1;
  72                                 }
  73                         }
  74                         else if(seqA[i] != '-' && seqB[i] != '-'){
  75                                 if(seqA[i] != seqB[i]){
  76                                         difference++;
  77                                 }
  78                                 minLength++;
  79                                 openGapA = 0;
  80                                 openGapB = 0;
  81                         }
  82                 }
  83
  84                 //non-overlapping sequences
  85                 if (!overlap) { minLength = 0; }
  86
  87                 if(minLength == 0)      {       dist = 1.0000;                                                  }
  88                 else                            {       dist = (double)difference / minLength;  }
  89         }
  90
  91 };
  92
  93 /**************************************************************************************************/
  94
  95 #endif
  96