[ssml] Re: Slides to demonstrate conservation of protein structure?

[Dan Bolser]

On Wed, 13 Oct 2004, Andrew Coulson wrote:

>Which point in particular? The one about structures migrating? I can't 
>remember when this occurred to me, or what triggered it -- may just have 
>been me slipping into Devils Advocate mode in face of some triumphalist 
>crystallographer...... Is there any way we could get any evidence for its 
>plausibility?


I think the 'transitive' alignments you describe would be a good way to
test the idea. You could carefully align a succession of sequences and
show while the sequence alignments can cope with certain geometric
alterations, traditional structural alignments cannot.

Any clear examples of conformational change in evolutionary time? I was
once hoping to study (search for) this type of thing, as I don't know of
any clear rules that can be give for when / if this happens. I never
looked into it though.

However, I think structure alignments based on fragments, and allowing
flexible regions (energy minimizing one alignment onto another) could
counter these problems with the existing techniques.

It is a good point you make though - our concepts are based on the tools
used to investigate the data. Of course nature doesn't give a damn about
conserving anything in particular, just increasing reproductive potential 
within the given physical laws of the universe.


P.S. Can you send the Kabsch & Sander refernce?

Cheers,
Dan.

>ac
>Or the whole thing? Yes, I've sometimes thought about planting a squib 
>somewhere -- wouldnt have thought it was up to Nature's august heights 
>though....
>
>At 12:29 13/10/2004, you wrote:
>>Andrew,
>>This is a really interesting point.  I shall change my lectures this year
>>because of it.  Its an important issue and would make a good short letter
>>to Nature or elsewhere,
>>James
>>
>> >Dan ---
>> >
>> >
>> >>Now I have another question - What is the original reference for the 'well
>> >>known fact' that protein structure is more conserved than sequence?
>> >         I'm glad you've put this question again, because it gives me a
>> >chance once again to clamber onto my oldest and most battered pedantic
>> >hobby-horse.
>> >This 'well-known fact' does not have a respectable provenance, because it
>> >is a meaningless assertion. Sequence conservation can be measured, say, by
>> >%identity; less intuitively, structure conservation may be measured by RMS
>> >deviation. These are qualitatively different scales, and asking whether a
>> >pair of sequences is more conserved than a pair of structures is like
>> >asking whether Frost is 'closer' to Larkin than Boston is to Hull. 'You
>> >can't add apples to oranges', as Mr Ellenger said on Day 1 of Algebra at my
>> >(English) prep school.
>> >The study by Chothia and Lesk which you refer to showed empirically that
>> >there is a (non-linear) mapping between the two measures (see for example
>> >Fig 5.6 on p181 of Lesk' 'Introduction to Protein Architecture'). But the
>> >whole point of such a mapping is to allow distances on one scale to be
>> >converted to distances on the other (or rather, to make approximate
>> >estimates of the conversion) -- so that the underlying 'evolutionary
>> >distance' between two proteins comes out the same whichever way you look
>> >at it.
>> >         It would be a meaningful -- and true -- assertion to say that
>> >evolutionary conservation may often be more easily recognised by comparing
>> >structures than by comparing sequences, but the inverse argument does not
>> >necessarily hold. If a pair of structures are recognisably similar, but the
>> >sequences are not, this may be because of convergent evolution. There are
>> >three types of evidence which argue for divergent evolution:
>> >1) (much the strongest) is where you can establish a chain of pairwise
>> >sequence similarities, each individually strong, but the ends of the chain
>> >have no sequence similarity -- for example, leghemoglobin and the
>> >oxygen-carrying globins.
>> >2) (a much used shortcut) is to use similarity of function, and perhaps
>> >conservation of active site residues, etc. This argument has been used, for
>> >example, to suggest that several TIM-barrels, whose function is saccharide
>> >hydrolysis, are evolutionarily related.
>> >Both of these arguments depend on evidence outside the naked structure;
>> >3) is to argue that so many features of the structures which might have
>> >been different are the same that chance and convergent evolution are
>> >improbable. These arguments can be problematic because of the uncertainties
>> >involved in 'which might have been different'. Many years ago, Rossmann
>> >argued that subtilisin had a common ancestor with the Rossmann-fold
>> >dehydrogenases because, inter alia, the handedness of the beta-alpha-beta
>> >connections was the same throughout. It was quickly realised that this
>> >handedness is energetically favourable.
>> >            Finally, can I make a pair of mischievous suggestions?
>> >One is that since we know (from Kabsch & Sander, initially) that identical
>> >subsequences may have differing structures, isn't it possible that
>> >occasionally 'sequence is more conserved than structure'?
>> >And the second is to wonder whether structure-based alignments are always a
>> >more reliable guide to evolution than sequence-based ones. Isn't it
>> >conceivable, for example, that a beta-strand might migrate through a
>> >sequence (steps of two-at-a-time, perhaps) after the insertion or deletion
>> >of loop residues? In such a case, structurally-aligned residues would not
>> >have a common ancestor.
>> >
>> >Andrew Coulson
>>
>>
>
>
>----------------------------------------------------------------rasmol-+
>
>To change your address, unsubscribe, view previous messages/history,
>receive messages as weekly digests, or for any other information
>about the RasMol EMail List, please go to
>        http://www.umass.edu/microbio/rasmol/raslist.htm
>
>----------------------------------------------------------------rasmol--
>