The Need for Local Alignments.

Suppose the two residue gene "WE" is washing around the primordial soup and quite by chance gets inserted into "QWERT" and "AWEPQ". Well, why not? So along we come, knowing that global alignments are of little interest because of this very attitude on the part of genes of pushing their way into various and sundry genetic parties. Well, the optimal global alignment, using the same primitive scoring scheme we've employed to this point, is:


That's not bad, but it doesn't focus attention where we want it. If we stop punishing end gaps, then the best alignment is:

- - - - Q W E R T
A W E P Q - - - -

That's pretty miserable. One simple way of seeing if there are any reasonable local alignments is the dot plot. Given two sequences we write one out horizontally, the other down vertically from the left edge of the first, and with this arrangement of letters we form a rectangular grid. (This will all become very obvious on the next page). Each position on the grid corresponds to one letter from each sequence. If the letters match, we color the position in; if not, then we leave it white. Then we look at the result and anywhere we see a diagonal string of colored positions, or even one or more such strings almost touching at the ends, then we can be pretty sure we've found something significant.

Well, let's do it. When you press the "Next" button you'll be taken to a page that will prompt you for two sequences. You can use any characters you like, but don't use more than forty. And the more you use, the longer it takes. After you've put in the second sequence and pressed the "ok" button you may have to wait for a minute or so for the JavaScript to do its thing. Be patient, then carry on.