Biology Reference
In-Depth Information
Fig. 1 Homologous over-extension—self-alignment of cortactin (
SRC8_HUMAN
) using the
lalign
program using BLOSUM62,
11/
1. Five significant alignments are produced in addition to the 100 %
identity alignment. Shown is the second most significant alignment, with E() < 10
29
, which aligns cortactin
domains 1-5 with 3-7
psisearch
that dramatically reduce over-extension are available at
the EMBL-EBI [
3
]. In proteins, alignment of relatively closely
related domains (
50 % identity) using a matrix like BLOSUM62
or BLOSUM50, which target alignments with 25-30 % identity,
can cause the alignment to extend well beyond the homologous
domain boundaries into non-homologous sequences (Fig.
1
).
Homologous over-extension is most easily recognized when
the average percent identity of the aligned sequences differs dra-
matically from one part of the alignment to another. For example,
Fig.
1
shows a local
lalign
self-alignment of cortactin (
SRC8_
HUMAN
) using BLOSUM62,
>
1, between cortactin domains
1 and 5 and homologous cortactin domains 3-7. Across the homol-
ogous regions (residue 80
11/
on the bottom,
highlighted in bold), the domains average 60 % identity. But the
first 80 residues of the alignment shown are non-homologous and
average less than 15 % identity.
When the
SRC8_HUMAN
is aligned with itself using
VT80
,a
scoring matrix appropriate for sequences that are 50 % identical, the
non-homologous over-extension part of the alignment disappears;
the alignment begins at residue 80 in the first sequence and residue
155 in the second. The protein scoring matrices available with
the FASTA programs make it possible to target the scoring matrix
to the evolutionary distance of the aligned sequences using the target
identity information in Table
7
. Using the appropriate matrix can
dramatically reduce homologous over-extension.
For very distantly related sequences, it is very difficult to identify
and correct homologous over-extension, and homologous “under-
extension”—an alignment that covers the most similar parts of two
on the top and 155
