"William R. Pearson" wrote:
I am in complete agreement with your comments. Absence of stuctural
similarity should be sufficient evidence to rule out homology. I was
attempting to point out that short internal homologies may exist in otherwise
dissimilar proteins; the evidence for this is seen in many motiffs (nucleoside
binding pockets, etc). I realize now that the trypsin and subtilisin similarity
was a poor example.
> To argue that the active sites of trypsin and subtilisin are
> homologous, is to argue for the inference of homology in the absence
> of any objective evidence. The active site of each protein is
> comprised of three residues, which are spaced throughout the length of
> each protein sequence. There is no signficant sequence or structural
> similarity shared by the two proteins; trypsin is comprised of two
> beta-barrels while subtilisin is a largely alpha structure.
>> There seem to be two uses of the term "homology". Some evolutionary
> biologists extend the meaning of "common ancestry" well beyond what
> can be tested experimentally, and may sometimes argue that all
> proteins are ultimately homologous, based on the assumption that there
> was a single primordial gene that, presumably through multiple
> duplication events, gave rise to all current genes.
>> I prefer a definition of homology that is based on some objective
> measure of similarity. My ex-graduate student, Todd Wood, pointed out
> to me that the original use of the work "homology" by Richard Owen,
> first curator of the British Museum in the 19th century, was based on
> morphological or developmental similarity. From this perspective, all
> genes (or proteins) cannot be homologous, because it makes no sense to
> talk about the homology of dissimilar structures, e.g. an all-beta
> protein and an all-alpha protein. One might argue that they shared a
> very ancient common ancestor, but it seems equally reasonable to
> believe that both structures arose independently. This perspective
> demands evidence for homology; the null hypothesis is that almost all
> genes are non-homologous.
>> The inference of homology, based on statistically significant sequence
> or structural similarity, is an exceptionally powerful and reliable
> tool for characterizing newly sequenced genomes. The concept of
> homology would be considerably less valuable if it could not be used
> to infer structural similarity.
>> Bill Pearson