next up previous contents index
Next: Similarity versus distance scores Up: Dynamic programming for sequence Previous: Variable gap penalty   Contents   Index

Local versus global alignment

The Kruskal and Sankoff version of the local alignment is implemented [Sankoff & Kruskal, 1983]; this is very similar to the [Smith & Waterman, 1981] method. All the routines for the local alignment are exactly the same as the routines for the global alignment except that during the construction of matrix $D$ the alignment is restarted each time the score becomes higher than a cutoff. The second difference is that the backtracking starts from the lowest element in the matrix, wherever it is.



Ben Webb 2005-06-20