A compressed alphabet uses one letter to represent a group of similar amino
acids. For example, F might be used for F, Y and W. The advantage of using a compressed
alphabet is that some of the most common substitutions (e.g., F for W or L for
V) are tolerated without losing the exact matches required for word
indexing. This improvement in sensitivity because matching words are found in
more distantly-related sequences. A significant (so far, mostly unpublished) research
effort was invested to determine the optimal alphabet and pattern for typical medium- to low-identity protein searches.
Compressed amino acid alphabets are supported for all search and clustering commands that
use indexes (see index
options). The alphabet can be specified using the -alpha option.
The alphabet is specified as a string containing every letter in
standard 20-letter amino acid alphabet exactly once. Groups of
letters that are to be represented by a single symbol in the
reduced alphabet are separated by commas.
The order the groups are specified is not
significant. By convention, the group is represented by its first
letter, so for example in the above alphabet the 4-mer MARW would
be compressed to IAKF.
By default, protein ublast is the only
command that uses compressed alphabets. This is because other
commands are generally used for high-identity search and clustering
tasks where compressed alphabets typically give only a marginal
improvement in sensitivity.
The default compressed
alphabet and pattern for protein ublast
could be specified by the following options:
Edgar, RC (2004)
Local homology recognition and distance measures in linear time
using compressed amino acid alphabets, NAR 32(1),