Lessons from the pandemic: A new look at an new virus: patterns of mutation accumulation in SARS-CoV-2 since 2019

That sentence is incoherent. You first say flexibility diminishes the correlation substantially, but then say “in those cases”. But a correlation would be a statement about the trend across all cases.

Second, a correlation is just that, a correlation. It is not my contention that there are no exceptions, and @Giltil has already acknowledged that the two are not the same. All he said was that there is a correlation and he’s cited an article that explains why and how there is. And that article isn’t some sort of fluke or outlier. It is a well-known and established phenomenon in the field.

It is of no use for you and @Mercer to then speak of those cases where the correlation doesn’t hold, as if to deny there is a correlation. And let me remind you that is what @Mercer did. He said no when @Giltil said the two are expected to correlate.

Well, they are.

Keefe and Szostak are aiming at finding what is the frequency of functional proteins in a random library. It is obvious that in their mind, functional proteins must have the attribute of specificity in addition to affinity. They made this point very clear right in the following passage in their introduction: « We selected for proteins that could bind a small molecule target with high affinity and specificity as a way of identifying amino-acid sequences that could form a three-dimensional folded state with a well-defined binding site and therefore exhibit an arbitrary specific function ».
Bottom line: Your quarrel with Rum and me over affinity vs. specificity is irrelevant in the context of the Szostak paper. It is also irrelevant in the context of natural antibodies, since the process of affinity maturation generates antibodies that are not only more affine and but also more specific.
https://www.cell.com/cell-reports/pdf/S2211-1247(19)31104-0.pdf

There is nothing incoherent about my response and it was meant to show the flaw in Gil’s earlier claim (in bold) that

There are cases, and a lot, that are enough to invalidate the use of the phrase “in general” in Gil’s claim above.

Obviously, no such correlation exists across all cases.

It is Gil’s contention that it is general and that is false!

I wasn’t denying a correlation existed at some level, I rejected it was the molecular norm in general as that is false.

No. Mercer was rejecting Gil’s claim of the existence of a general correlation, just as me. Refresh your memory:

The article he cited didn’t show the existence of this correlation “in general”.

You seem to be forgetting that Gil stipulated high affinity and high specificity, explicitly not all cases. I noted that the correlation deteriorates when one tries to maximize either parameter:

So, do you agree or not?

Gil also clearly stipulated “high” a second time:

So why are you pretending that I made any global claim?

You left out high affinity again. Why?

No, it’s highly relevant, because as you even admitted:

So using their frequency in any comparison is all but meaningless without stipulating the affinity threshold they used.

That’s why your comparison was ridiculous:

So, Keefe and Szostak’s highest-affinity ATP-binding protein had an apparent Kd of 200 nM. To even begin to make any reasonable comparison with the antibody frequency, you’d have to know what typical Kds of antibodies from naïve B-cells are. What is that number? I’ll also note that a thousand-fold difference is not as huge as you are portraying it to be. In fact, when biochemists discuss binding, they tend to use factors of 1000 as the first demarcation. Hypothetical conversation:

A) “So we found this great candidate in our screen that binds really well!”
B) “How well? Nanomolar?”
C) “No, picomolar!”
D) “Wow!”

Keefe and Szostak’s frequency is meaningful enough to demolish Doug Axe’s extrapolation, because no adjustment for affinity is ever going to change anything by a factor of 10^66.

Can we at least agree on that?

Yes, particluarly since they undergo negative selection for self binding. There’s a lot going on there.

But you were claiming that most of the information is present in the genomic VDJ segments BEFORE maturation, remember? The reality is that so much information is created de novo in two weeks, that it demolishes the rotting edifice of ID.

I think you misunderstood me. To say that affinity and specificity correlate in general (across all cases) is to say that we expect them to follow each other more often than not if we were to take an average of the relationship across all cases, and that when one goes up then it is more likely than not that so does the other (if it was less often than not, or in general if one goes up then the other goes down, then they would either be uncorrelated or anti-correlated.)

But such a correlation does exist, and it is well-known. It is observed for anything from small molecule ligand binders, protein-protein or protein-DNA/RNA binders, to the activity and specificity of enzymes.
In general (more often than not), the more specific the binding or activity, the greater the activity and/or binding also is. In general. Not always, but in general.

That isn’t a statement that there are no exceptions where, for some reason, one does come at the cost of the other. But in general (if you take an average across all cases), it is true that they track each other for the reasons well explained in the paper cited by @Giltil.

That means were we to pick some protein sequence at random and select it for more specificity towards some ligand over some other ligand, we would expect (the most likely outcome) that to also result in more affinity towards the same ligand. And we would expect that to happen more often than we would expect them to uncorrelate or anti-correlate were we to do the same experiment with another protein towards another target.

This is what I am saying. I believe this is what @Giltil is saying, and it is what the paper he referenced argues. This phenomenon is well known and, frankly, the only place I have ever seen it disputed is here by you and @Mercer. For reasons that make no damned sense to me.

With what specifically? I agree affinity and specificity often go together. More often than not, which is why it is thought they generally correlate such that you expect if you select for an increase in one, the other will be expected to also increase.

I’m not sure whether I agree that the correlation should weaken at the extremes of specificity and affinity. At least I will say I don’t know the degree to which that effect might be real. I can see how in principle one can come at a cost of the other, but I don’t see why that effect should be more pronounced at the tail ends of high specificity or affinity.

Obviously I agree that we should be careful about what kinds of extrapolations and inferences we make from that and what consequences they can have. But in a sort of generalized argument about what we expect to be most likely outcomes from hypothetical cases of protein evolution, the most reasonable position is to take well-known and established positive correlations seriously, rather than to dismiss them as if they’re not even real.

When you and @Michael_Okoko argue that the correlation doesn’t even exist, as you have done, then you’re just not being reasonable.

Yes, this is exactly what I am saying!

The reason is that some people here have sadly adopted the principle that nothing should be conceded to an ID proponent, even when he says perfectly reasonable things.

Stop with the straw man. We’re not talking about all cases. You got in a lather over my comments objecting to:

See that word “high”?

See the word “high” again?

See the word “high” again?

Do you see the word “all” in any of those?

Did you qualify your objection with “high”?

This was never about all cases. It’s about the bleeding edge.

We weren’t discussing all cases.

We weren’t discussing in general.

Why do you keep insisting that we were discussing something we weren’t?

It isn’t.

That’s because you put words in our mouths to get lathered up about.

Then maybe you should learn about it instead of pretending that Gil and I were arguing about ALL cases. Gil stipulated HIGH affinity. I stipulated HIGH affinity.

Then maybe you should spend less time getting cranked up over straw men and more time reading before responding?

You’re not behaving as though you do.

Get a grip. You’re manufacturing this.

Thanks for the clarification. That was helpful.

That correlation appears to only exist for fairly rigid molecular recognition systems. When binding involves recognition systems with significant flexibility, that correlation disappears, as more often than not, a high specificity is associated with low or moderate affinity (and sometimes vice versa as some authors argue). Interestingly, this decoupling of specificity and affinity is physiologically vital. In addition, flexible molecular recognition is widespread in vivo, so the issue of it being an exception should not arise as it certainly isn’t.

Show me data.

False, in the context of flexible molecular recognition (and it is pervasive).

Considering that intrinsically disordered proteins and proteins with intrinsically disorder regions are abundant in nature, and they are well known for exhibiting flexible molecular recognition (hence, decoupling affinity and specificity more often than not), this shouldn’t work.

Of course, good data will change my mind.

Thank you.

I agree that the exception to the rule is conformational flexibility. If a molecule can change shape to match another target, then it is necessarily more lacking in specificity as this very ability allows it to adopt the shape of numerous different potential targets, and yet this flexibility can also potentially enable great affinity at least because it can very closely match the shape of the target.

However, I am not making a claim about the distribution of flexible vs relatively rigid molecules found in living organisms. Nor am I making a claim about what is most useful or adaptive to some given organism. I know that conformational flexibility is a widespread phenomenon, and
yes I absolutely agree with both @Mercer and you that molecules with these properties are vital to cellular life as we know it.

My claim is about both what we expect would most likely happen if we imagine picking sequences randomly and blindly from sequence space and start selecting for either affinity or specificity(we would expect selection for one to also increase the other as a byproduct), and about what we know from directed evolution studies of ligand binding proteins and enzymes, and even from studies of affinity maturation of antibodies.

Sadly I can’t find a sort of review or metaanalysis of the phenomenon (there might be one, I just don’t know how to find it), so I’d have to go and find you all the different studies on directed evolution of ligand binders, enzyme activity etc. to “show you the data”. But I concede that’s a poor substitute for a systematic review.

While looking I did however find various articles that support @Mercer’s statement that maximizing activity or affinity comes at a cost of specificity, and maximizing specificity comes with a cost in terms of reduction of affinity or activity. It’s difficult to find articles that show how widespread these phenomena are, though.

2 Likes

From the standpoint of antibody diversification, this underplays the significance of randomization and especially maturation which is acknowledged throughout the relevant literature. In any event, the point skirted is that antibodies are complementary to antitopes. In covid, we have witnessed the typical pathogen behavior which is to vary antigenic sites to evade existing immunity defenses. The immune system may react to any number of viral proteins, bacteria, fungi, venoms, foreign tissue, parasites, pollen, food, and generally unknown unknowns. Thus, the information specifying all these real or perceived threats, out there at large in the wild, is very wide. If these are the range of distinct safes to crack, the complementary information cannot pre-exist in a much more limited storage. I am not persuaded that information specificity for omicron variants spike protein pre-exists in gene segments.

The discussion of the correlation of affinity and specificity highlights an essential aspect of antibody information, which is that a given antibody operates as a whole and not as the sum of its parts. Affinity and especially specificity are outcomes of an integral shape and would not be the same given some other arrangement. Alternative antibodies do not represent more or less information, but different information which define different characteristics. One unique antibody for one unique specification.

I do not believe that you have addressed the significant objection which others have raised to your statement, in that nature does not generally have to find complementary shape spaces from scratch in any event. Quite the opposite; modification and re-application of existing features is typical. Therefore, the random reassortment and mutation which takes place in antibody development is not materially distinct from random reassortment and mutation which drives evolutionary development. The ID argument that evolution is unable to generate new complementary shapes does conflict with the immune system routinely employing the same conceptual process to find complementary shapes to antitopes.

Another way to put it is if the information for specific antibodies exists in the unmodified V(D)J region of the genome then the information needed for every species alive today existed in the last universal common ancestor since all subsequent species were just a rearrangement, modification, and duplication of that original genome.

1 Like

I invite you to have a look at the piece below by Gpuccio (as well as his response to critics) regarding the process of affinity maturation of antibodies. As he says, the interesting point is that the whole process, which has been defined as “darwinian”, is in fact the best known example of functional protein engineering embedded in a complex biological system.

https://uncommondescent.com/intelligent-design/antibody-affinity-maturation-as-an-engineering-process-and-other-things/

The problem here is that it is highly misleading to claim that the immune system operates according to the same conceptual process than evolution (cf @gpuccio’s piece).

Yet another argument from definition. All too often, ID proponents think that if they call something engineered or designed that the mere declaration is somehow evidence that they are engineered or designed. In nearly every case, they fail to address the basic mechanisms at play, which is exactly what you have done in your post. You are more concerned with what things are called instead of how they actually work.

2 Likes

That’s because it is.
A Darwinian process has the following four components:

  1. A stochastic process that generates variation in blindness to it’s phenotypic effect.
  2. More variants generated than can actually survive.
  3. The variation is heritable.
  4. The phenotypic effects of the heritable variation affects the probability of survival and reproduction so there is competition between variants.

If a process has those components it is Darwinian. How does that stack up against affinity maturation?

  1. Somatic hypermutation and VDJ recombination are both stochastic, and done in blindness to their phenotypic effects.
  2. More b-cells presenting antibody variants are generated than go on to survive.
  3. The variation in the genes encoding the antibodies is heritable.
  4. The phenotypic effects of the heritable variation affects the probability of survival and reproduction of the different b-cells, so there is competition between them.

It has the components of and therefore meets the defining characteristics of a Darwinian process then.

1 Like

No, it is not. The affinity maturation of antibodies is an example of intelligent selection, not natural selection. I let you with the confutation given by @gpuccio to DNA-Jock who was wandering how the selection was intelligent:

“How is this selection “intelligent”?”

« I use the term “intelligent selection” to differentiate it from “natural selection”.

In “NS”, the variation confers a reproductive advantage, and therefore it is “selected” (the term selection here is not completely proper), in the sense that it expands in the population and is more or less fixed (positive and negative selection).

The key point here is that the variation itself confers an advantage to the organism in terms of fitness and reproduction. There is no special organized system in the environment that “measures” any property of the variation. In a sense, the new organism “selects itself” because it can use better the resources of the environment.

IS is any situation in which the system actively measures some property of the mutated object and reacts to that measure in a specific way. In this case, only the measured property can trigger the active response of the environment which “selects” (this time in a completely correct sense) the result, and expands and supports it.

In affinity maturation, the process is an intelligent selection in all senses. The variations in sequence due to random (but controlled) mutagenesis generate different affinity for the hapten which was originally used to select the responsive clones. The hapten is retained in the Follicular GC cell, and then is used again to measure the variation of affinity. Two different active responses of the system are triggered by the measure (indeed, we still don’t know, as far as I can say, how that part works).

In the case of a reduction of affinity, apoptosis in initiated. That is in no way a direct consequence of the higher affinity of lower affinity of the new molecule, because the B cell at present is in no way engaged in defending the organism from any attack. It is the programmed response to a specific signal, to the result of a measure.

On the other hand, if the affinity is higher, the B cell is “protected” by the intervention of specific T helper cells. Again, this is a programmed response to a specific signal (the higher affinity, measured against the retained hapten), and has no immediate advantage, except in view of what can happen in the future (a new encounter with the invading microorganism). Again, it is an intelligent response to the interpretation of a signal.

I am not implying here that any active intelligence is working in the cells. The process is embedded as an algorithm, and it can go on because it was programmed to do so, exactly like a computer program will go on even if it is not “actively intelligent” (IOWs, conscious and cognitive) in that moment.

But, as a computer program performs passively intelligent algorithms, so does the biological system which generates the affinity maturation.

In no way there is a “natural selection” here. And if you say that the environment selects like in darwinian natural selection, because after all the environment includes an antigen presenting cell, a measure system, a system which actively generates controlled variation in a specific part of the gene, a system to destroy the bad results (apoptosis) and a system to promote the good results (specific T helper cells), and that we can consider all that exactly like a fitness function in a natural system, well, then I will be very disappointed by you. »

Rarely does one see so many words say so little that amounts to nothing more than just blatant denial.

In the case of a reduction of affinity, apoptosis in initiated. That is in no way a direct consequence of the higher affinity of lower affinity of the new molecule

So is it or is it not a consequence of the higher or lower affinity? It is. Let me help Gpuccio make up his mind here. It is a consequence of the affinity. The affinity directly translates to the probability of successful conjugation with TFH cells.

because the B cell at present is in no way engaged in defending the organism from any attack. It is the programmed response to a specific signal, to the result of a measure.*

Who gives a crap if the b cell is involved in defending the organism from any attack? There isn’t any part of natural selection that says the basis of selection has to involve “defending the organism from any attack”. Completely made up and silly qualification.

It’s actually explained well on wikipedia of all places: Affinity maturation - Wikipedia

Clonal selection: B cells that have undergone SHM must compete for limiting growth resources, including the availability of antigen and paracrine signals from TFH cells. The follicular dendritic cells (FDCs) of the germinal centers present antigen to the B cells, and the B cell progeny with the highest affinities for antigen, having gained a competitive advantage, are favored for positive selection leading to their survival. Positive selection is based on steady cross-talk between TFH cells and their cognate antigen presenting GC B cell. Because a limited number of TFH cells reside in the germinal center, only highly competitive B cells stably conjugate with TFH cells and thus receive T cell-dependent survival signals. B cell progeny that have undergone SHM, but bind antigen with lower affinity will be out-competed, and be deleted. Over several rounds of selection, the resultant secreted antibodies produced will have effectively increased affinities for antigen.

It’s selection based on the differences in the phenotypic effect of mutants, it affects the reproductive success of carriers through their competition for a limited resource, and it’s a natural process based directly, entirely, and only on the physics and chemistry of molecular interactions. Natural selection in textbook form.

It’s a Darwinian process. Get over it.

In no way there is a “natural selection” here. And if you say that the environment selects like in darwinian natural selection, because after all the environment includes an antigen presenting cell, a measure system, a system which actively generates controlled variation in a specific part of the gene, a system to destroy the bad results (apoptosis) and a system to promote the good results (specific T helper cells), and that we can consider all that exactly like a fitness function in a natural system, well, then I will be very disappointed by you. »

Gpuccio’s disappointment with me will be hard to live with. But let me take this opportunity to state that natural selection always involves factors that determine and cause carriers of heritable variation to show differential survival and reproductive succes. Whether that happens to be antigen presenting cells, or lions suited for hunting slow antelopes, or wind suited for carrying pollen. Those are physical things in the environment that determine the survival and reproductive success of different carriers of variation.

LOL

3 Likes

Not to mention the selective advantage of having our improving such a system.

This is an unexcusable, strategic omission of a system that amplifies, in a perfectly Darwinian way, a lot of bad results…

Do you know what that system is, Gil?

It affects the reproductive success of carriers because a whole complex system is there to to the job through a measurement and then a selection of a property that is not naturally tied to reproductive success. So, no, it is not natural selection. It is not a Darwinian process. It is intelligent selection. The difference is best explained by Gpuccio below:

1) Natural selection: in a system, reproducing beings compete for the system’s resources. RV generates better reproducers, which expand, suppressing (more or less) the previous population.

2) Intelligent selection: we have a system with reproducing beings. Those beings have properties which are not directly connected to reproduction. The system has complex configurations that can measure one of those properties. Controlled RV changes the degree of that property. The system measures the variation. The system can react with two different behaviours to the measure: one of them suppresses the reproduction of the reproducing beings, the other one enhances it. The system implements one or the other according to the binary result of the measure. Therefore, the reproduction of the reproducers is suppressed or enhanced according to the variation of the property, even if the property itself gives no reproductive advantage.

In the second type of process, the connection between the property and the differential reproduction is guaranteed by a complex configuration of the system, which implements an algorithm with different steps and logical connections.

Your iPhone operates also entitrely through the natural laws of physics and chemistry. It doesn’t mean it can entirely be explained by natural law.