Variants of the dynamic programming algorithm
As an example for the different alignment variants consider the comparison of estrogen receptor and some other receptor.
Alignment Quiz: 4.
There is an interesting interplay between parameters,
in particular the gap penalty, and the algorithmic variant used.
Consider a pair of sequences whose similar regions can in
principle be strung together into an alignment
(as opposed to sequences containing repeats which cannot
all be seen in a single alignment).
Under a weak gap penalty the Smith-Waterman algorithm has a chance
to identify this entire alignment. On the other hand, not knowing
about the similarity between the sequences ahead of time,
a weak gap penalty might also cause all kinds of spurious aligned regions.
The Waterman-Eggert algorithm is a valid alternative.
The gap penalty can be chosen fairly stringently.
The first (i.e., the Smith-Waterman) alignment will then
identify only the best matching region out of all the similar regions.
By iterating the procedure, though, this algorithm will successively identify
the other similar regions as well.
Comments are very welcome.