Optima in Evolution

That’s a silly attempt to shift the evidentiary burden.

The point is that “FI” doesn’t detect other optima. The point is that YOU don’t know, yet you claim to.


There may be many functional alternative solutions, but that doesn’t make them all local optima. It actually appears that for ATP synthase no such inescapable “local optimum” has been found, as the sequences of basically all it’s subunits have been diverging for all cellular organisms, continuously for the entire known history of life.

No, that isnt’t the reason I am contesting his calculation. I am contesting his calculation because the method isn’t capable of determining how many functional sequences there are.

It’s not merely that I think he’s wrong that makes me contest his calculation. I am contesting the calculation primarily because:

  1. The method is based on a sampling process that has extremely high intrinsic bias.
  2. The fraction of sequence space the process samples is an infinitesimally small one.

His method is not capable of telling us how many functional sequences there are. For reasons explained now so many times it boggles the mind anyone would have still have trouble understanding it.

As stated before, even if Gpuccio’s number is actually correct(which there is no good reason to think), the method he uses is not how that is established, and he would only happen to be coincidentally correct in the number. Sort of like if he had happened to correctly guess the true number of functional sequences, and it was later found to be corroborated in some elaborate experiment, that would not mean that just guessing is how we determine the FI for some protein sequence. No, we really do have to use a method that is capable of telling us that, and for reasons given above many times, sequence conservation used to derive FI isn’t how that is done.

Functional sequences, not just local optima. But yes, how many functional sequences are there? You can’t extract that information from picking extant sequences from living organisms and extrapolating from that.

So an invisible, telepathic, palestinian Columbidae, from classical antiquity, instead of Santa, must have made it?

In all seriousness, you seem to have this weird picture in mind where there was a not-yet-fully-formed ATP synthase molecular machine, that lacked subunit beta of the F1 hexamer, and evolution had to sort of “find” a functional beta-subunit before the system would function. It’s the Hoyle-fallacy all over again.
But of course that’s not how these things happen. The individual proteins that make up ATP synthase have ancestries that go back to times before they became part of the ATP synthase complex. The real question is how frequent is the ultimate ancestral function of the entire superfamily to which the beta subunit belongs?

The beta subunit did not evolve de novo into the ATP synthase complex, rather it belongs to a gigantic superfamily of proteins called P-Loop NTPases(which, ironically, also contains the actin and myosin subfamilies), one of the most divergent and oldest protein superfamilies known. The beta subunit is part of the catalytic hexamer(the one that also contains the Walker-A motif), which is actually functional on it’s own as a DNA and RNA translocase(and as an RNA helicase).

The most basic function of all P-loop NTPase members is nucleotide binding (and hydrolysis). That is the inferred ancestral function and the essential role of them all.
That’s the function they tested for in the Keefe & Szostak 2001 experiment (ATP binding), and found to exist at a frequency of approximately 10^-11(noteworthy here is that they found four different completely dissimilar proteins with the function, which would have been undetectable as being related in any BLAST or other similarity-based search). Later experiments revealed that under certain conditions, the protein also catalyzes ATP hydrolysis to ADP.
A version of the original experiment was repeated by another lab, and results published in 2015. This time they used a smaller subset (15) of the 20 known biological amino acids, biased towards the set of amino acids thought to be prebiotic, and found additional proteins capable of ATP binding and hydrolysis.

The Tawfik lab constructed a super simplified facsimile of the inferred ancestor of the P-loop NTPases, called the Walker-A P-loop motif. The protein is only 55 amino acids long, yet functional, in that it could bind and hydrolyse ATP(and other NTPs), bind RNA and DNA.

That function really does seem to be hyperabundant in protein sequence space.


Yep. Gil doesn’t seem able to shift out of this “Gpuccio gives me this number, prove him wrong or he’s right by default”-mode of thinking.

1 Like

Just more “But this one goes to 11.”

Predictably, Gil hasn’t explained why “FI” isn’t perfectly analogous to Nigel’s amplifier knob.

While on the subject of P-loop NTPases, the extreme functional diversity of that superfamily of proteins shows that functional protein sequence space is actually highly interconnected, and different functions are overlapping. The same protein sequence and structure can harbor many different functions, and functions can be incrementally derived from each other with relative ease by mutations and natural selection.

1 Like

You are missing the mark here for I was not talking about the whole ATP synthase but only the beta chain that happens to exhibit a high degree of conservation from bacteria to human. So why are you trying to muddy the waters here by invoking other subunits?

You appear to misunderstand me, I was not speaking about the whole complex. What I wrote is true for the beta subunit considered in isolation. It really is the case that it is highly divergent across the tree of life, and that different ATP synthases have different beta subunit sequences.

And not overlapping too. The myosins, the kinesins, and the G proteins are all P-loop NTPases. The same protein fold at the core.

This is why @Agauger’s going on about “functional folds” made no sense.

1 Like

Related: I just read about this

”To kick-start the i5k, a pilot project was launched at the Baylor College of Medicine led by Stephen Richards to sequence, assemble, and annotate the genomes of 28 diverse arthropod species carefully selected from 787 community nominations.“

“ The team’s analyses focused on tracing gene evolutionary histories to estimate changes in gene content and gene structure over 500 million years. This enabled identification of families of genes that have substantially increased or decreased in size, or newly emerged or disappeared, or rearranged their protein domains, between and within each of the major arthropod subgroups. The gene families found to be most dynamically changing encode proteins involved in functions linked to digestion, chemical defence, and the building and remodelling of chitin – a major part of arthropod exoskeletons.”

“ We can take these hypotheses into the lab and use them to directly study how the genome is translated into visible morphology at a resolution that cannot be achieved with any other animal group,” says co-lead author, Ariel Chipman, from the Hebrew University of Jerusalem, Israel.

Santa Clause has more evidence for existence that does ID-Creationism. :slightly_smiling_face:

1 Like

I predict that a Gish Gallop is forthcoming…

What wouldn’t have a marginal impact on FI?

1 Like

The level of polymorphism

What level of polymorphism would be a problem for FI, and why?

1 Like

After reflection, I can see that I’m not quite sure to understand your question. Can you precise what you mean?

What would have a major impact on FI relative to ID?

1 Like

I guess that if it could be demonstrated that there exist a huge number of alternative solutions in the sequence space for implementing complex biological functions, then this would destroy the design inference based on FI.

Isn’t that exactly what the 1,000’s of neutral missense mutations in the human MYH7 gene demonstrates?

1 Like

That is of course built on the never shown to be true, and actually demonstrated-to-be-false, assumption that FI can be estimated based on extant variation observed in living organisms.

To make matters worse, even if it could be shown that some functions really do have very high FI (like 500 bits or more), that still doesn’t imply they couldn’t evolve.

The whole argument from the ID camp is an a priori belief, seemingly based on nothing but hunches that functional proteins must be rare in sequence space, and that different functions with different frequencies have little to no overlap in sequence space. Both of these a priori beliefs have no evidence in support of them.

Concrete real-world experiments reveal that functional proteins with adaptively beneficial phenotypic effects, are in fact selectable from random sequence libraries, and can even evolve de novo from non-coding DNA. And that such proteins can have multiple functions in the same sequence, can be converted incrementally from one function to another by mutations and selection, and that their beneficial functions can be enhanced gradually by natural selection acting on mutations.

This is also strongly implied by the entire field of comparative genetics, which shows that there have been a relatively steady influx of new functional proteins over the history of life in basically all clades on the tree of life. Both in terms of divergence through duplication and mutation of ancestral proteins, and by de novo evolution from non-coding DNA.
A curious person has to wonder, if such new functional proteins really are supposed to be so rare as to be practically impossible to evolve, why one would get such a result from using phylogenetic methods?

There is a strange kind of anti-realism about how ID creationists deal with such patterns seen in comparative genetics.

Instead of thinking that they might be wrong about their a priori beliefs about how rare and impossible de novo(or divergent) protein evolution is, they seem instead to go straight to the inference that the result from phylogenetics must mean a designer has to have intervened many times during life’s history to create these novel proteins in different clades. So in this case the a priori belief takes precedence over real-world evidence, is held constant as if an unassailable axiom, and the data is interpreted through the lens of the a priori axiom, rather than letting the evidence guide them into what to believe.

It’s sort of like having an a priori belief that a coin that was once flipped was a fair coin no matter what. Then the ID creationist comes upon a large collection of recorded coin flips, sees that the coin has landed tails 90% of the time. Then the ID creationist concludes an unobserved invisible force, an intelligent coinflipper, must have made the (still assumed to be fair) coin land tails, instead of inferring that maybe the coin just wasn’t fair.


No, not at all, a first reason being that the large majority of these 1,000’s of neutral missense mutations in the human MYH7 gene would not correspond to positions that exhibit conservation through deep time.