A Response to David Gelernter’s Attack on Evolution

There are no specified changes in evolution. Neither human speech, bird calls or vertebrate eyes were specified in advance, and none of them had to evolve.

Your conclusions are based on a false premise.

3 Likes

Yep. It’s the same “sharpshooter” logical fallacy ID-Creationists have been tripping over forever. The claim what we see was somehow targeted and is the only possible combination which will support life.

1 Like

Maybe for once you could try actually discussing the rebuttal scientific evidence instead of knee-jerk defending anything to do with Creationism.

1 Like

So, @bjmiller, in the paper you point to, exactly how many “specific, coordinated mutations” are described?

3 Likes

Heres Bechly and company:

“What about the other Ediacaran trace fossils? All gone. A seminal study published in 2016 experimentally demonstrated that these Ediacaran trace fossils can be easily reproduced as artifacts of stirred up bacterial mats that covered the Ediacaran sea floors.”

Here’s the author of that study discussing this paper: https://royalsocietypublishing.org/doi/10.1098/rsos.172250

“Yes, that looks like an animal burrowing into the sediments

The mechanism I describe in my article would definitely not explain that.

Cheers

G”

Soooo yeah. Looks like all the Ediacaran trace fossils aren’t gone after all.

@bjmiller it isn’t hard to email these guys and make sure you got them right…

3 Likes

Some have argued that catalytic antibody (abzyme) research challenges the argument for extreme protein rarity. In reality, the truth is the exact opposite. As a case study, Shahsavarian et al. used a phage display library of the size on the order of billions to generate catalytic antibodies approaching the efficiency of beta-lactamase enzymes in breaking down antibiotic molecules. Antibodies are highly specialized multicomponent proteins that are designed to maintain a stable structure as localized sections of the protein known as Fv regions dramatically vary. One Fv region resides at the end of each of the antibody’s two branches, and it consists of the variable domains within a heavy chain and a light chain. The immunoglobulin gene randomizes the variable regions allowing for a binding site to eventually appear that can bind to a target and possibly break it apart. Finding the right combination of amino acids to degrade an antibiotic molecule proved relatively easy.

Yet, abzymes function very differently from enzymes. In the former, the variable domains forming a binding site consist of localized sequences of amino acids held in fairly consistent positions by nonvarying sections known as constant regions. The constant regions also ensure the variable regions in the heavy and light chains reside at the right locations in close proximity. In contrast, an enzyme starts off with the amino acids which form the catalytic site residing at distant locations along the chain. The folding process forms the active site by moving the correct amino acids to the right locations and positioning them in the right orientations.

Moreover, in enzymes both the active site and amino acids throughout the protein structure are specified to assist in its target function. Specifically, an enzyme’s entire conformation morphs into multiple configurations. This complex dynamic is well summarized by Hammes et al.,

Multiple intermediates, multiple conformations, and cooperative conformational changes are shown to be an essential part of virtually all enzyme mechanisms.

Each reconfiguring involves the coordinated rearrangements of single amino acids and often entire secondary structures.

Therefore, the tasks of forming a functional abzyme and generating a novel functional enzyme represent fundamentally different problems. A new enzyme requires both finding a set of amino acids with the right chemical capacities and generating a new fold that brings those amino acids together properly in 3D space and provides structural support. The fold also must perform complex conformational changes to support specific chemical activities. The abzyme only needs to stumble upon the correct amino acid sequences in the variable regions for the catalytic activity. The amino acids are already positioned properly by the constant regions, and the latter also provide the needed structural support. In addition, abyzmes do not morph their overall conformations to assist specific chemical reactions. These differences explain abzymes’ limited capacities, and they result in enzymes having much greater functional sequence rarity.

Ironically, the abzyme research greatly strengthens the argument for the generality of extreme rarity, for it shows that degrading antibiotic molecules is a relatively easy function to achieve. In contrast, the enzyme HisA participates in an intermediate step in the synthesis of the amino acid histidine where it performs a highly specific molecular rearrangement. Namely, the enzyme detaches a hydrogen atom from one nitrogen molecule and attaches a hydrogen atom to another nitrogen. No abzyme or polypeptide generated in a randomized library has ever demonstrated a comparable ability to reengineer molecules.

The difference in the difficulty of antibiotic degradation and molecular reengineering explains beta-lactamase’s greater resilience to accumulating mutations than HisA’s. This difference directly translates into HisA’s more extreme sequence rarity. Many enzymes and structural proteins also perform more difficult tasks with greater specificity requirements than beta-lactamase, so a 10% populated target region should be an optimistic estimate for a large percentage of globular proteins.

There are those interesting moments where you read one sentence and can almost 100% guarantee that everything after that sentence will be wrong.

There are plenty of examples of standard enzymes where the residues responsible for catalytic activity are in a contiguous segment.

This is also false. There are plenty of enzymes where chunks of the protein can be removed without affecting activity. Also, “specified” is a meaningless term.

No, they don’t. In the case of abzymes, there is a ~10 amino acid section of the protein that is randomized. This is no different than the insertion of a random sequence (e.g. random recombination event) into a non-antibody protein.

Sharpshooter fallacy. You are painting the bulls eye around the bullet holes. Evolution isn’t trying to hit a target. All evolution is doing is selecting for genetic changes that increase fitness.

It is entirely possible that there would be additional enzyme functions if the variable region were located elsewhere in the protein. Again, you are committing the sharpshooter fallacy.

Perhaps you should tell Douglas Axe that.

B-lactamases take atoms from water and re-engineers the b-lactam with those atoms.

2 Likes

But you never go on to explain why. You just wrote a huge wall of text that contains exactly zero valid reasons for reaching that conclusion.

So only sixty eight orders of magnitude higher than estimated by Axe.

Shouldn’t they have needed to generate 10^77 sequences before they found a beta-lactamase function?

But I’ve read on EN&V that it should only happen roughly 1 in 10^77 attempts.

But now you’re telling me relatively little evolutionary change is needed to discover a biologically useful function, such as catalyzing the breakdown of a specific antibiotic molecule?

Funny.

Uhm, no. The fact that it is “easy” to degrade antibiotic molecules does not at all strengthen the argument for the extreme rarity of functions. In any way. Straight up non-sequitur.

It seems to me to imply the diametrically opposite. If sequences with different but useful functions were all extremely are in sequence space, and isolated from each other (as they would need to be to prevent evolution from discovering new functions from already existing functional sequences by sampling into their immediate surroundings), then why is it so easy to turn an antibody protein into an enzyme with a biologically useful function? Shouldn’t those two be both incomprehensibly rare, and totally isolated from each other?

No, it doesn’t.

Every time it comes to supplying the evidence or reasoning that is purported to support your conclusion, it turns out the statement you make is either just flat out false, or at best a mere blind assertion. Or implies the diametrically opposite of the ID narrative.

3 Likes

One must be very careful to distinguish between what some might imagine being true from what has been demonstrated to be true empirically. Dan Tawfik has stated that all theories about the origin of new protein folds are based primarily on speculation. Ignoring the implications of hard data by appealing to speculative theories is like a trial lawyer ignoring video footage of a crime in favor of the testimony of his client’s imaginary friends.

The hard evidence points to the following:

  • Enzymes can only evolve to the extent that the structure does not change, the active site remains basically the same, and the catalyzed chemistry is similar. Tawfik labeled these changes as micro-transitions.
  • Evolving a new protein fold requires an evolving gene to pass through regions of sequence space without any function.
  • A straightforward mathematical analysis of studies on the effect of random mutations on protein stability/function demonstrates that sequences corresponding to functional proteins are exceedingly rare.

The analysis of protein rarity is now much more accessible to the public. Doug Axe’s 2004 JMB article was extremely difficult to understand by anyone who was not an expert in the field. Consequently, critiques of his work could use erroneous arguments, and the public was powerless to identify the errors:

In contrast, Tawfik’s experiments can much more easily be interpreted. For instance, roughly half of all beta-lactamase mutants with three random amino acid changes are still functional. That change corresponds to a 1% alteration in the initial sequence. And, nearly all mutants with 10% of the sequence randomly altered are nonfunctional. In comparison, a 10% change in the letters of a short paragraph is still largely readable. Therefore, functional protein sequences are rarer than readable English paragraphs.

In addition, a large proportion of proteins consist of combinations of a limited number of domains just as a limited number of words are used in most sentences. This pattern was described by Scaiewicz and Levitt, and they identified numerous other similarities between protein sequences and human language including syntax, semantics, grammar, and the importance of context.

This observation relates to the common error of claiming that estimates of protein rarity exaggerate the difficulty of finding a functional target since other proteins or other distinct versions of the same protein might exist which could perform the same function. A multitude of alternative targets could dramatically increase the odds of finding one of them. Yet, this possibility seems remote given the extremely low probability of a random search entering a target region. It is also challenged by the fact that newly discovered multidomain proteins are very often “combinations of domains characterized by a limited number of sequence profiles.” If sequence space contained such vast numbers of targets, newly discovered proteins should not repeat the same sequence and structural patterns so often.

In addition, the bacteria population exceeds the population of most eukaryotic taxa by many orders of magnitude (e.g. 10^30 bacteria verses 10^13 trees). Yet, the percentage of taxonomically restricted genes (TRG) in ash trees is 25%, to name just one example, and this percentage is at least as large as the percentage in most taxa of bacteria. Some have argued that the TRG estimates are greatly exaggerated due to limited sampling, but a recent paper from Carvunis’ lab challenges this argument. The fact that TRG numbers in bacteria do not vastly outnumber those in eukaryotic species also strongly suggests that sequence space is not supersaturated with targets.

Um, beta lactamase is a hydrolase. There are more than just a few hydrolases in the biosphere, and their substrates are many and extremely varied.

So @bjmiller, how many papers can you cite in which someone has screened a random combinatorial library for a reaction that resembles the HisA reaction? If the answer is zero, then your point is meaningless.

3 Likes

@bjmiller, no need to explain why you don’t understand Axe’s work, or the criticisms that working biochemists have made, criticisms that are valid and unanswered by Axe or anyone else.

Ask, and we can take you through this.

2 Likes

You ask a very dangerous question. This is just the sort of question that led David Gelernter and Gunter Bechly to lose their neo-Darwinian faith. The Durrett and Schmidt article calculated that the time required for 2 specific, coordinated mutations to appear in humans or whales is longer than the time in which they were believed to have evolved from their hypothetical ancestors. The article also calculates the time for two specific, coordinated mutations, which are both neutral, to appear and spread throughout a Drosophila population. That time is around 9 million years. The authors relate two mutations to the appearance of a new transcription factor binding site.

We can apply these results to the origination of vocalization. Different bird species capable of vocalization (repeating sounds) possess the same amino acid substitutions, unseen in other birds, in 6 specific genes associated with the vocalization-dependent brain regions. And, 66 genes show specialized expression. In addition, bird vocalization shares many similarities with human vocalization in both neural architecture and genetics. For instance, a few hundred genes demonstrate shared specialized expression in birds and humans.

Yet, these changes are just the tip of the iceberg. Vocalization also requires several specialized regions of the brain with numerous highly targeted neural connections. One non-local connection requires the highly coordinated production of chemical signals to guide a neuron’s axon to the proper location to form a functional circuit. This coordination must correspond to several new transcription factor binding sites.

Equally striking, echolocation in one group of bats and dolphins involves the same 21 aa changes in Cdh23 and 22 aa changes in Pcdh15. And, the implementation of echolocation and other aquatic-mammal traits involves several other innovations:

Needless to say, before vocalization, echolocation, or other novel innovations emerge at even the most rudimentary level, far more than two specific, coordinated mutations would be required.

Compounding the problem, no mutation has ever been observed which has added a single neural connection to any animal which enhanced some ability. Nor have any been observed which have made any significant change that could drive a major transformation.

The argument of whether evolution faces severe mathematical challenges is analogous to two astronauts finding the remains of a pole vaulter on the moon and then pondering if the athlete could have traveled there under his own power. The first astronaut might look at the world record for successful pole vaults and then compare that number to the distance to the moon. The natural conclusion would be that the difference in distances is too great to justify the possibility. The second might retort by saying that many factors could increase a vaulter’s potential maximum height such as nutritional supplements, better training, and better poles. He could then state that one simply could not be certain whether vaulting to the moon might be possible.

The same difference of opinion resides with evolution’s potential. Many feel the math demonstrates, beyond a reasonable doubt, that an undirected process could never generate the novel innovations seen in life in the available timeframes. For instance, the appearance of new phyla in the Cambrian explosion requires countless highly coordinated changes at the foundational level of an animal’s architecture in a timeframe which would only account for the appearance and fixation of a single specific pair of neutral mutations. Given such a disparity, is questioning evolution’s limitless potential truly unreasonable?

I’m a bone and stone guy so my knowledge isnt going to be as great as the other guys on this thread but I just don’t get this “specific mutation” argument. Evolution isn’t looking for a specific mutation(s). And we know there are many pathways to the same function. So evolution doesn’t have to find “specific mutations” because there are many who would do the job just as good as the two “specific mutations”.

I think of papers like this:

3 Likes

Excellent points and well said. love the ehcolocation data and how math does make the mutation odds thing impossible to create biology’s glory. the bird vocals was new to me. As a YEC creationist i do insist marine mammals changed bodyplans to go from land to sea. Yet not by mutations being selected in a crazy impossible way. i do see biology having other mechanism to change bodyplans as we see in people.
i do tell folks that its just a parts department number from a common blueprint that better explains the likeness in genetic info. thus the echolocation thing is case in point.

So you concede that your claim that one protein superfamily needs to be changed incrementally into another protein superfamily by point mutations is a straw-man.

Ignoring the implications of hard data

That’s you doing that. It’s you ignoring hard data. Like the Tawfik paper you read but didn’t understand. And the talk you quote from but didn’t watch or understand.

What is your hard evidence for that?

The Tawfik video you linked above argues that enzymes can evolve to catalyze radically different chemistries and accomodate very different substrates, so your last point there seems to be completely opposite to demonstrable fact.

  • Evolving a new protein fold requires an evolving gene to pass through regions of sequence space without any function.

Evolving a “new protein fold” from what? A nonfunctional sequence, another functional one? And what exactly do you mean by a “new protein fold”? Do you mean change a specific protein from one structurally defined protein superfamily, into another specific protein from another structurally defined protein superfamily? Why would such a transition need to occur in the first place? What is your “hard evidence” for that?

Your continued insistence that such a transition must occur is without merit.

  • A straightforward mathematical analysis of studies on the effect of random mutations on protein stability/function demonstrates that sequences corresponding to functional proteins are exceedingly rare.

So rare that they are routinely found in screens of random sequence libraries, and are continuously evolving de novo. So much for whatever “straightforward mathematical analysis” you’re referring to here.

You don’t actually know that. You haven’t tested them for all possible functions, or under all possible environmental circumstances. And the changes you’re referring to here are the absence of purifying selection. You just keep “forgetting” this.

In comparison, a 10% change in the letters of a short paragraph is still largely readable. Therefore, functional protein sequences are rarer than readable English paragraphs.

That doesn’t actually follow.

You don’t show it to be an “error”, common or not. And you don’t show how it “relates” to it at all. You just blithely declare these two things with zero support.

And you’ve casually moved from talking about protein “folds” (which it’s still not clear what you mean by), to functions. Which are not the same thing.

A multitude of alternative targets could dramatically increase the odds of finding one of them. Yet, this possibility seems remote given the extremely low probability of a random search entering a target region.

But you haven’t shown that the probability is extremely low. In fact, the opposite seems to be the case. For example finding proteins capable of degrading antibiotics is apparently easy.

It is also challenged by the fact that newly discovered multidomain proteins are very often “combinations of domains characterized by a limited number of sequence profiles.”

How so? All that tells you is that multidomain proteins evolve by fusing existing proteins. What the hell does that have to do with how rare functional proteins are in sequence space? Nothing, it’s got nothing to do with it.

If sequence space contained such vast numbers of targets, newly discovered proteins should not repeat the same sequence and structural patterns so often.

Why? You just assert this but offer zero reasons for it.

“Bacteria verses trees”? What? Just what the hell are you talking about? The paper talks about biomass. What does that have to do with the number of genes?

Yet, the percentage of taxonomically restricted genes (TRG) in ash trees is 25% to name just one example

That reference doesn’t actually show that.

Some have argued that the TRG estimates are greatly exaggerated due to limited sampling, but a recent paper from Carvunis’ lab challenges this argument.

These claims are so vague as to be meaningless. Who have argued that TRG estimates are “greatly exaggerated”?[citation needed] How exaggerated did they argue them to be?[citation needed] And to what extend does that reference “challenge” it? If others have argued the numbers are exaggerated , then to what extend does this paper argue that the numbers are not exaggerated?

Also, if the number of TRG that evolve is high, doesn’t that show that it’s easy to evolve new functional proteins? Just what the hell are you even trying to argue here?

The fact that TRG numbers in bacteria do not vastly outnumber those in eukaryotic species also strongly suggests that sequence space is not supersaturated with targets.

You have shown zero data that compares the number of taxonomically restricted genes for bacteria vs eukaryotes. And the mechanisms that would cause de novo gene birth in eukaryotes and bacteria are not identical and do not occur with the same frequency, so such a direct comparison would be completely meaningless.

I have to concede that I was impressed by how much Gish-galloping nonsense you managed to mash into that last paragraph.

3 Likes

It’s only dangerous because the answer disagrees with your characterization of the study you cited. Specifically, the answer could range from 1 to some sum of the numbers you list further in your response to my question. (If you don’t understand how this statement is correct, feel free to ask.)

In the 10+ years since the Durrett and Schmidt article came out, how many other papers have been published that expand on (or correct) their assertions? What (if any exists) is the consensus in the field? @bjmiller, surely you know that Durrett and Schmidt are not the last (or first) word on this subject.

We’re not talking about an “argument” and we’re not talking about “extreme protein rarity.” We’re talking about the ratio of protein sequences that are functional to the total number of protein sequences.

Here’s an example of you misrepresenting the data. While the library was 2.7 billion, they had 5 hits, meaning that the ratio (what you’re allegedly, but not really, addressing) is on the order of 1 in only 500 million.

If Doug Axe’s extrapolation from lousy binary data had any validity, they would have never found any. Note that Axe didn’t measure enzymatic activity. Why didn’t he, Brian?

You’re misusing both “designed” and “stable” here. Are you saying that if the H and L V regions were entirely positively-charged residues, that the C regions would still maintain a “stable structure,” whatever that means? Is more stability always associated with better function?

Therefore, Axe’s sloppy extrapolation from an enzyme with the identical activity is wrong.

There’s no contrast. Only some do, and even worse for you, abzymes and many other enzymes have the amino-acid residues that form the catalytic site on separate subunits!

Gee, apparently you missed this paper:
https://www.pnas.org/content/93/11/5590.long

Now, given that Doug Axe, the first author, wrote this:
Of the active mutants produced, several have no wild-type [of 13] core residues. These results imply that hydrophobicity is nearly a sufficient criterion for the construction of a functional core and, in conjunction with previous studies, that refinement of a crudely functional core entails more stringent sequence constraints than does the initial attainment of crude core function.”

How can you make the obviously and objectively false claim that “amino acids throughout the protein structure are specified,” Brian? Are you really not familiar with Axe’s own work, or are you cynically cherry picking?

Abzymes do not require “generating a new fold.” They are made up of repeats of the immunoglobulin fold, a very common fold. The immunoglobulin structure brings those amino-acid residues, which are on two different proteins, together properly in 3D space and provides structural support.

No difference there. You clearly don’t know what you are talking about.

How much grant money will the DI bet? You’ve stumbled onto a testable hypothesis that I’ll bet none of you have the slightest interest in testing.

I don’t see any contrast there. Why would you use the segue “in contrast” when there’s no contrast?

Again, how much money would you (and the DI) like to bet on your claim? Do you have faith in it?

We don’t need to “evolve a new protein fold.” A single fold has many different functions and different folds can have the same function. A single protein, functioning normally, can adopt two different folds–that’s how many proteins work. It’s a structural classification, not a functional one.

The first is demonstrated well in this paper:
https://www.pnas.org/content/107/32/14384

Note that the authors use the correct term, “protein fold.”

3 Likes

It all comes back to this ludicrous idea of pre-specification.

I’m pretty sure that the average waiting time for the origin and fixation of two particular pre-specified mutations is a meaningless calculation in the context of particular evolutionary transitions. Evolution isn’t (and weren’t) searching for those two mutations. It’s searching for any adaptive mutations. So what we really need to know is how often adaptive double mutants could be expected to occur on average.

The rate of fixation of neutral alleles is equal to the rate of mutation, which would be somewhere in the range of 100 to 150 every generation in the human population.

So it seems to me the real question is, on average, how many of those 100-150 singularly-neutral mutations are adaptive in conjunction? That would then tell you how many adaptive double mutants which are neutral alone, that you could expect to establish in a population over some given number of generations.

3 Likes

can it be that most of them are adaptive?