Calcium-Binding Protein Protocols Calcium-Binding Protein Protocols

246 Weljie and Heringa
much larger length of sequences, which provides the opportunity for a greater
number of consensus residues in the C2 family. This leads to many more gapless
columns than found with the EF-hand alignments, which enhances the sampling
during the bootstrapping calculations. Finally, both alignments resulted in phylogenetic
trees that appropriately classify the sequences into subfamilies (see
website).
3. Annexin proteins: The annexins constitute a heavily studied protein family from
a phylogenetic point of view (e.g., refs. 11 and 12) partially because of their
important biological function. Another reason stems from the fact that the annexin
repeat is a reasonably common eukaryotic motif; however, the number of
orthologs and paralogs found provides a considerable challenge in classification
because of the lack of defined identity between proteins from the same species. For
example, 10 human annexins have a mean amino acid identity of 49.8 ±/– 4.1% (12),
and estimates of mutational rates suggest that between very different species (e.g.,
plants and animals) the identity should be much lower. Paradoxically, the annexin
repeat itself is particularly interesting because the sequences are all highly
homologous, similar in length with essentially no gaps or deletions, lending themselves
well to alignment. The alignment of the annexin proteins by PRALINE
and CLUSTALX are given in Fig. 3. The degree of similarity in terms of sequence
length is immediately striking, and on further analysis the conservation of key
residues is also dramatic. The NJ trees created with these alignments showed
reasonable stability on bootstrapping; however, neither alignment lends itself well
to classification of the annexins into appropriate subfamilies. Presumably such
analysis is better suited for alignments based on the entire annexin protein, and
not simply on the repeat itself. It should be noted that the most complete classification
of subfamilies to date has been accomplished through a combination of
DNA and protein analysis, in conjunction with maximum-likelihood methods for
phylogenetic analysis (11).
4.3. Multiple Alignment Method
The problem of finding an optimal or highest scoring alignment of two
sequences was solved three decades ago with the DP technique (39), which
guarantees the finding of the highest scoring alignment determined from summing
amino acid substitution scores minus any insertion/deletion penalties.
The amino substitution weights are normally given as a 20 × 20 matrix, containing
the weights for all possible amino acid exchanges. The insertion/deletion
penalties are used to decrease the alignment score when gaps need to be
made to optimally match the two sequences. Normally, a pair of gap penalties
is used, consisting of an opening penalty used once for each gap and an extension
penalty applied to each incurring gap position. However, when applied to
more than two sequences, the calculation of the optimal alignment by multidimensional
implementations of the basic dynamic programming algorithm for
sequence pairs (64), becomes computationally unfeasible. Even with localized