On 1/30/2023 7:04 PM, Ernest Major wrote:
Commentary from Derek Lowe on research on making proteins with low
sequence similarity to natural ones - something which has relevance to estimates of the efficacy of evolution over the long term.
https://www.science.org/content/blog-post/making-proteins
Nature hasn't had to do this (make totally unique proteins that do the
same function). As noted only a very small fraction of sequence space
has had to be accessed by nature. It looks like most new genes are due
to duplication or recombination between existing sequences. There are a
bunch of basic motif sequences that serve some structural function, and
they get used in a bunch of different sequences over the course of
biological evolution. There are novel genes evolving out of random
sequence, but it looks like nature can select for something that works
(folds up efficiently) and then slight alterations can make the sequence
do different things.
My example has always been abzymes where you trick an antibody into
evolving an enzymatic function by creating an epitope that looks like a structural intermediate. Antibody function is to evolve a sequence that
can efficiently bind to an epitope (antigen). Mutations are induced
into the existing antibody sequence and tested out, and it turns out
that a very small fraction of protein sequence space has to be tested by modifying this existing sequence to produce new enzymatic function.
Less than 2 trillion sequences have to be tested. That number is an
over estimate and comes from making the obviously claim that fewer cells containing unique antibody sequences have to be tested and selected over
a 2 week period than the number of cells in the mouses body.
So by taking a sequence that already folds efficiently you can make
fewer than 2 trillion unique sequences by mutating that existing
sequence and create a new function. The variable region of the antibody
is only around 130 amino acids in length. There is the intelligent
design nonsense estimate that the chance of assembling a 100 amino acid sequence has the low probability of something like 1 in 10^191, and yet
using an existing protein sequence our antibody system can put in
arbitrary mutations and select a product with a new enzymatic function
in less than 2 X 10^12 tries. Only a very small fraction of the
possible sequence space needed to be searched to identify the new
functional sequence.
This seems to be how the protein genes evolved on this planet. Sequence
that worked got duplicated and repurposed to do all that was needed to
evolve what we have today, and only a small fraction of protein sequence
space was required to do it. This is what the author is acknowledging
at the beginning of the article. More divergent sequences can do the
same thing, but there was never a requirement to create those more
divergent sequences when something similar to what already existed could
be changed to do the job.
Much more of the sequence space could have been intelligently
manipulated to create the same functional proteins, but that wasn't
needed in the history of life on earth.
Ron Okimoto
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)