[ Previous | Up | Next ]

As we mentioned above, global sequence alignment algorithms align sequences over their entire lengths. You do need to think about whether that type of alignment makes sense for your sequences. For our example, where we expect each exon to be represented in the sequences and in the same order, it has worked well - however, how well do you think this approach would work with, for example, multidomain proteins that share one domain but not others, or sequences where there have been regions of duplication? A second comparison method, local alignment, searches for regions of local similarity and need not include the entire length of the sequences. Local alignment methods are very useful for scanning databases or when you do not know that the sequences are similar over their entire lengths. The EMBOSS program water is a rigorous implementation of the Smith Waterman algorithm for local alignments.

EMBOSS contains other pairwise alignment programs - stretcher and matcher are global and local alignment programs respectively that are less rigorous than needle and water, and therefore run more quickly; they may be useful for database searching. Supermatcher is designed for local alignments of very large sequences and is even less rigorous in its implementation. A comparable program to supermatcher would be megablast.

Exercise 5 - Global and local alignments: needle and water

Use the applications needle and water to align human globin mRNA to bovine globin mRNA Do not adjust the gap parameters from the defaults used in these programs.

How do the results differ? Look both at the statistics and the alignments themselves. How do these differences illustrate the difference between global and local alignments?

[ Previous | Up | Next ]

Page last modified September 29, 2008