My group has been using the profile server in developing a very large multiple alignment of aldehyde dehydrogenases. The alignment now contains over 120 sequences, including some proteins of marginal catalytic function such as lens crystallins. Particularly in finding the most distantly related sequences the Profile server has been invaluable. We are also in the process of searching for examples from other families that might contain a similar NAD-binding site, by using truncated alignments to enhance the chances of finding such domains. (It is now known that ALDHs have non-traditional Rossmann folds). Otherwise, a side project involving an alignment of arginases also has made use of the Profile server. Both projects are manuscripts in preparation. What we did in that project was refute a hypothesis based on aligning single ADH and ALDH sequences with a protein of about the length of the 2 combined, all from Aspergillus. To eyes unfamiliar with the characteristics of ADHs and ALDHs, the hypothesis was not too unreasonable, but what previous au's failed to appreciate (since they never read any relevant literature!) was that none of their matching residues were among those typically conserved in those 2 families. What we did was run profiles of each family based on 16 ALDHs and about 100 ADHs, and run the profile against the big Aspergillus sequence. The result basically said no similarity, but that if the big protein did represent an ancestral fusion of the 2, they were fused in opposite order from the earlier proposal, thereby refuting the whole earlier hypothesis on a nice impartial, statistical basis!
Showing the most recent 10 out of 270 publications
