Aphrodite’s head: Eight questions for Douglas Axe

Over at Evolution News, Dr. Douglas Axe argues that merely by using very simple math, we can be absolutely certain that life was designed: it’s an inescapable conclusion. To illustrate his case, he uses the example of a rugged block of marble being transformed by natural weather processes into a statue of a human being. Everyone would agree that this simply can’t happen. And our conclusion wouldn’t change, even if we (i) generously allowed lots and lots of time for the statue to form; (ii) let each body part have a (discrete or continuous) range of permitted forms, or shapes, instead of just one permitted shape; (iii) relaxed the requirement that all body parts have to form simultaneously or in sync, and allowed the different parts of the statue to form at their own different rates; and (iv) removed the requirement that the different parts have to each form independently of one another, and allowed the formation of one part of the statue to influence that of another part.

In his post, Axe rhetorically asks: if we’re so sure that a rugged block of marble could never be transformed by the weather into a human statue, then aren’t we equally entitled to conclude that “blind natural causes” could never have “converted primitive bacterial life into oaks and ostriches and orangutans”? In each case, argues Axe, the underlying logic is the same: when calculating the probability of a scenario which requires many unlikely things to happen, small fractions multiplied by the dozens always result in exceedingly small fractions, and an event which is fantastically improbable can safely be regarded as physically impossible.

In an attempt to persuade Dr. Axe that his logic is faulty on several grounds, I’d like to put eight questions to Dr. Axe, and I sincerely hope that he will be gracious enough to reply.

My first question relates to the size and age of the universe. As I understand it, Dr. Axe, you define “fantastically improbable” as follows: something which is so improbable that its realization can only be expected to occur in a universe which is much bigger (or much older) than our own. Indeed, on page 282 of your book, Undeniable, you further stipulate that “fantastically improbable” refers to any probability that falls below 1 in 10116, which you calculate to be the maximal number of atomic-scale physical events that could have occurred during the 14-billion-year history of the universe. You calculation requires a knowledge of the age of the universe (14 billion years), the amount of time it takes for light to traverse the width of an atom, and the number of atoms in the universe. So here’s my first question for Dr. Axe: how is the design intuition supposed to work for an ordinary layperson who knows none of these things? Such a person will have no idea whether to set the bar at one in a million, one in a billion, one in 10116 , or even one in (10116)116. I should also point out that the figure you use for the number of atoms in the universe refers only to the observable universe. Astronomers still don’t know whether the size of the universe as a whole is finite or infinite.  And it gets worse if we go back a few decades, in the history of astronomy. Until the 1960s, the Steady State Theory of the universe was a viable option, and many astronomers believed the universe to be infinitely old. How would you have argued for the design intuition back then? 

My second question relates to functional coherence. You make a big deal of this in your book, Undeniable, where you managed to distill the case for Intelligent Design into a single sentence: “Functional coherence makes accidental invention fantastically improbable and hence physically impossible” (p. 160), where functional coherence is defined as a hierarchical arrangement of parts contributing in a coordinated way to the production of a high-level function (p. 144). The problem with your statue illustration should now be apparent. A statue has no functions. It just sits there. Consequently, whatever grounds we may have for rejecting the supposition that ordinary meteorological processes could transform a block of marble into a statue, they obviously have nothing to do with the argument you develop in your book, relating to functional coherence and whether living things could possibly be the product of unguided natural processes. So my question is: will you concede that the marble block is a bad illustration for your argument relating to functional coherence?

My third question relates to the identity of the object undergoing transformation. In your statue illustration, you ask whether “a rugged outcrop of marble would have to be altered by weather in only a few reasonably probable respects in order to convert it into a sculpted masterpiece.” Obviously, the answer is no: the number of steps would be extremely large, and the steps involved would be fantastically improbable. You then compare this case with the evolutionary claim that “blind natural causes converted primitive bacterial life into oaks and ostriches and orangutans.” But there is an obvious difference in the second case: the primordial bacterium itself is not being changed into an orangutan. Its very distant embryonic descendant, living about four billion years later, is developing into an orangutan. Its ancestors 20 million years ago were not yet orangutans. Self-replication, along with rare copying mistakes (mutations), is required in order for evolution to work. So I’d like to ask: why do you think it’s valid to infer from the fact that A’s changing into B is a fantastically improbable event, that A’s distant descendants gradually mutating into B is also fantastically improbable?

My fourth question relates to chemistry. Let me return to your original example of a block of marble being transformed by weather events into a human statue. I think we can all agree that’s a fantastically improbable event. However, the probability is not zero. I can think of another event whose probability is much, much lower: the likelihood of weather processes transforming a block of diamond, of adamantine hardness, into a human statue. What’s the moral of the story? Chemistry matters a lot, when you’re calculating probabilities. But the average layperson, whom you suppose to be capable of drawing a design inference when it comes to living things, knows nothing about the chemistry of living things, beyond the simple fact that they contain atoms of carbon and a few other elements, arranged in interesting structures. An ordinary person would be unable to describe the chemical properties of the DNA double helix, for instance, even if their life depended on it. So my question to you is: why do you think that a valid design inference can be made, without knowing anything about their underlying chemistry?

My fifth question relates to thermodynamics. I’d like you to have a look at the head of Aphrodite, below (image courtesy of Eric Gaba), known as the Kaufmann head. It’s made of coarse-grained marble from Asia Minor, and it dates back to about 150 B.C.

You’ll notice that her face has worn away quite a bit, thanks to the natural weather processes of weathering and erosion. This is hardly surprising: indeed, one might see weathering and erosion as an everyday manifestation of the Second Law of Thermodynamics: in an isolated system, concentrated energy disperses over time. Living things possess an unusual ability to locally decrease entropy within their
highly organized bodies as they continually build and maintain them, while at the same time increasing the entropy of their surroundings by expending energy, some of which is converted into heat. In so doing, they also increase the total entropy of the universe. But the point I want to make here is that a living thing’s highly useful ability to locally decrease entropy is one which a block of marble lacks: its thermodynamic properties are very different. So my question to you is: why would you even attempt to draw an inference about the transformations which living things are capable of over time, based on your observations of what happens to blocks of marble? And why would you encourage others to do the same?

My sixth question relates to your probability calculations. In your post, you explain the reasoning you employ, in order to justify a design inference: “it takes only a modest list of modestly improbable requirements for success to be beyond the reach of chance.” You continue: “Once again, the reasoning here is that small fractions multiplied by the dozens always result in exceedingly small fractions.” Now, this kind of reasoning makes perfect sense, if we are talking about dozens of improbable independent events: all you need to do is multiply the probability of each event, in order to obtain the probability of the combination of events. But if the events are not independent, then you cannot proceed in this fashion. Putting it mathematically: let us consider two events, A and B. If these events are independent, then P(AB) is equal to P(A) times P(B), and if both individual probabilities are low, then we can infer that P(AB) will be very low: one in a million time one in a million equals one in a trillion, for instance. But if A and B are inter-dependent, then all we can say about P(AB) is that it is equal to P(A) times P(B|A), and the latter probability may not be low at all. Consequently, in an inter-dependent system comprising dozens of events, we should not simply multiply the small probability of each event in order to compute the combined probability of all the events occurring together. That would be unduly pessimistic. And yet in your post, you attempt to do just that, despite your earlier statement: “Do I assume each aspect [of the statue] is strictly independent of the others in its formation? No.” So I’d like to ask: if you’re willing to grant that the even the formation of one aspect of a statue may depend on the formation of other aspects, thereby invalidating the method of calculating the probability of the forming the whole statue by multiplying dozens of “small fractions,” then why do you apply this invalid methodology to the formation of living things?

My seventh question relates to the vast number of possible pathways leading to the formation of a particular kind of living thing (such as an orangutan) from a primordial ancestor, and the even vaster number of possible pathways leading to the formation of some kind of living thing from the primordial ancestor. The point I want to make here is a simple one: this or that evolutionary pathway leading to an orangutan may be vanishingly improbable, yet if we consider the vast ensemble of possible pathways leading to an orangutan, the probability of at least one of them being traversed may not be so improbable. And even if we were to agree (for argument’s sake) that the likelihood of an orangutan evolving from the primordial ancestor is vanishingly low, when we consider the potentially infinite variety of all possible life-forms, the likelihood of evolutionary processes hitting on one or more of these life-forms may turn out to be quite high. It is this likelihood which one would need to calculate, in order to discredit the notion that all life on earth is the product of unguided evolutionary processes. Calculating this likelihood, however, is bound to be a very tricky process, and I doubt whether there’s a scientist alive today who’d have even the remotest idea of how to perform such a calculation. So my question is: what makes you think that an untutored layperson, with no training in probability theory, is up to the task? And if the average layperson isn’t up to it, then why should they trust their intuition that organisms were designed?

My eighth and final question relates to algorithms. Scientific observation tells us that every living thing, without exception, is put together by some kind of biological algorithm: a sequence of steps leading to the formation of an individual of this or that species. The algorithm can thus be viewed as a kind of recipe. (Contrast this with your illustration of a statue being formed by blind meteorological processes, which bears little or no relevance to the way in which a living thing is generated: obviously, there’s no recipe in the wind and the rain; nor is there any in the block of marble.) In order for “blind natural processes” (as you call them) to transform a bacterial ancestor into an orangutan, the algorithm (or recipe) for making an ancient bacterial life-form needs to be modified, over the course of time, into an recipe for making an orangutan. Can that happen?

At first blush, it appears fantastically unlikely, for two reasons. First, one might argue that any significant alteration of a recipe would result in an unstable hodgepodge that’s “neither fish nor fowl” as the saying goes – in other words, a non-viable life-form. However, this intuition rests on a false equivalence between human recipes and biological recipes: while the former are composed of letters which need to be arranged into meaningful words, whose sequence of words has to conform to the rules of syntax, as well as making sense at the semantic level, so that it is able to express a meaningful proposition, the recipes found in living things aren’t put together in this fashion. Living things are made of molecules, not words. What bio-molecules have to do is fit together well and react in the appropriate way, under the appropriate circumstances. Living things don’t have to mean anything; they simply have to function. Consequently, the recipes which generate living things are capable of a high degree of modification, so long as the ensembles they produce are still able to function as organisms. (An additional reason why the recipes found in living things can withstand substantial modification is that the DNA found in living organisms contains a high degree of built-in redundancy.)

Second, it might be argued that since the number of steps required to transform a bacterial ancestor into an orangutan would be very large, the probability of nature successfully completing such a transformation would have to be fantastically low: something could easily go wrong along the way. But while the emergence of an orangutan would doubtless appear vanishingly improbable to a hypothetical observer from Alpha Centauri visiting Earth four billion years ago, it might not seem at all improbable, if the Alpha Centaurian also knew exactly what kinds of environmental changes would befall the Earth over the next four billion years. The probability of evolution traversing the path that leads to orangutans might then appear quite high, notwithstanding the billions of steps involved, given a suitably complete background knowledge of the transformations that the Earth itself would undergo during that period. In reality, however, such a computation will never be technically feasible: firstly, because we’d probably need a computer bigger than the cosmos to perform the calculation; and second, because we’ll never have the detailed knowledge of Earth’s geological history that would be required to do such a calculation. So my concluding question to you is: given that the probability of nature generating an orangutan from a bacterial ancestor over a four-billion-year time period is radically uncomputable, why should we trust any intuitive estimate of the probability which is based on nothing more than someone eyeballing a present-day bacterium and a present-day orangutan? 

Over to you, Dr. Axe. Cheers.

311 thoughts on “Aphrodite’s head: Eight questions for Douglas Axe

  1. phoodoo: But wind also makes lots of things, so what do you mean when you say the sort of things ?

    Humans make lots of stonehenge?

    No, but stonehenge has many indications that it was made by humans, besides merely that the stones have some particular arrangement. Carvings, toolmarks, and it’s structure conforming to certain celestial positions are all things that fit into our understanding of the evolution of human culture. Humans have built temples, sacrificial altars, arenas, observatories and ritualistic gathering sites, to the sun, moon and stars, to the seasons, and to animals, for tens of thousands of years.

    If you think you have a method of working out whether some particular stone, or it’s position relative to other stones in the environment, was designed, based on probability, then you can impress me by applying your calculation to the stones in this picture and show which one was designed (and one of them was):

  2. colewd:

    The alpha helix is the simplest secondary structure so the result is a conservative approximation of random mutation creating a secondary structure.

    We’ve been over this before. Most amino acids can take part in an alpha helix, with varying degrees of propensity. That’s not to say any random sequence will generate a helix of course; it depends on the neighbourhood; the overall effect of particular runs of sequence. But, they can be tuned. As with the overall hydrophobic moment I’ve mentioned before, the conformation of the helix itself can be ‘improved’ by serial substitution. Each substitution changes the landscape for neighbours – both sequential neighbours and those slightly more distant ones brought into physical proximity by a turn of the helix.

    The other thing you ignore, and I think it’s a biggie, is the role of tandem repeat. You only need a dozen or so residues and you’ve got nearly 4 turns of a helix. It would clearly be a mistake to look at an 8 turn helix and conjure up some probabilistic nonsense relating to getting the full length in a single leap, and ignore the possibility of end joining a duplicate. And of course such a mechanism repeated could in principle generate substantial lengths in short order.

    These two issues together render probability calculations intractable, IMO. It’s not like generating Shakespeare out of Alphabetti Spaghetti.

  3. colewd:
    Rumraket,

    Yes, the causes of mutations are understood.Whats not understood is what the mutations can generate.

    Sure it is. Mutations can change DNA sequences, which can control how proteins are expressed, or the amino acid sequences of proteins. That’s actually quite well understood.

    Plate tectonics forming a mountain and random change and selection forming a sequence are two very different problems.

    Yes, which is why the causes thought to create them are different.

    Mutations + selection -> change one DNA sequence into another.

    Plate tectonics + erosion -> change the shape of a large rock into another.

    It is mystifying how this can be so difficult a concept for some people.

    I can easily see plate tectonics forming a mountain what I cant see is plate tectonics forming a living organism or any mechanism driven by random change.

    I don’t see what the problem is. How do we get from the ancestral sequence to the present one? By accumulation of mutations. If sequene A differs from sequence B in position 23 by having adenosine in place of a guanosine, then we infer that a mutation caused substitution to happen, for example. What’s the big problem here?
    How is this not immediately obvious? How did this sequence come about? It mutated incrementally from some ancestral sequence, and some of them were favored by natural selection. What is it about this you have difficulty envisioning?

    Your gene duplication idea requires a similar gene to form. There are over 2000 unique gene families.

    So what? Where is the problem?

    With all due respect Rum this is not a strong argument. Again, you are confusing the event with the cause when you compare the probabilities.

    How so? I have not confused the “event” with the cause at all.

    -The earth moving and erosion creating a mountain is a non controversial hypothesis.
    -Random change plus determinism (gene duplication etc) creating functional information is.

    It is irrelevant whether is controversial. Of course I know it is controversial to you, that is why we are having this discussion in the first place. That is not itself some sort of argument.
    If “someone finds it controversial” was a valid argument, then no question at all would be settled. There are young Earth creationists who find plate tectonics controversial. Some people believe the Earth is flat, or that virgins girls can give birth to males. “I don’t believe it” is not an argument.

    “I don’t know how to refute an incredulous stare.” – David Lewis

    If your whole argument amounts only to “I just don’t believe it”, then just say that so we don’t have to keep arguing about it.

    I am observing a very special mountain :-).Could tectonics and erosion be the cause…. likely yes.

    Where is the calculation that shows this? Please show by calculation how this would be different from the mutation+selection creating some particular gene sequence from an ancestral one.

    I am observing an organism filled with functional sequences.Could random change plus determinism be the cause….likely no.

    Where is your calculation Bill? Please do the calculation for both entities. Produce the calculation for some particular rock or mountain, and the calculation for a particular protein.

    The cause is reproduction a highly deterministic process.

    And…?

  4. Rumraket,

    Where is your calculation Bill? Please do the calculation for both entities. Produce the calculation for some particular rock or mountain, and the calculation for a particular protein.

    Do I need a calculation to determine that you are the result of sexual reproduction?

    Do I need a calculation to determine gravity caused a pen to fall to the floor?

    I do need a calculation to determine if a “random” drop of coins caused 20 pennies on a table to come up heads.

    Identified deterministic process don’t need probability calculations to determine cause.

  5. Allan Miller,

    These two issues together render probability calculations intractable, IMO. It’s not like generating Shakespeare out of Alphabetti Spaghetti.

    I agree with this point. Do you agree with Jock’s assessment that 6 AA’s randomly selected have a 31% chance of forming an alpha helix?

  6. Rumraket,

    Rum, we can model moving plates creating a mountainous structure. We cannot model random change and determinism creating a multicellular organism.

  7. colewd:
    Allan Miller,

    I agree with this point.Do you agree with Jock’s assessment that 6 AA’s randomly selected have a 31% chance of forming an alpha helix?

    I don’t know, I didn’t see this claim. Nonetheless, when I scanned back I saw you reference this then try to extrapolate it as a ‘random pick’ to longer strings, which is exactly what I’m arguing against. Don’t ignore duplication, nor the tuning of near misses. Probabilistic approaches, from Hoyle on, invent difficulties for a randomising mechanism no-one proposes, then ignore all alternative mechanisms.

  8. Allan Miller,

    Don’t worry. That’s not a claim I ever made.
    I hereby challenge colewd to produce the original quote.
    There’s an extra word in there that he dropped,
    probably because he did not understand Chou-Fasman.
    ETA: your point about multiplication is absolutely correct, too.

  9. colewd:
    The alpha helix is the simplest secondary structure so the result is a conservative approximation of random mutation creating a secondary structure.

    No it isn’t such an approximation because it fails to consider possibilities other than alpha-helix. As I told you, what good is the math if you don’t know how to apply it properly?

    Even before considering evolutionary phenomena, we know that any random combination of amino-acids will have propensities towards one or another secondary structure. There’s three main ones: “loops,” alpha-helix, and beta-strands. Thus, we should expect a 2500 aa-long protein to contain combinations of these these main kinds of structures. What do you think the protein has? Yes! Segments of alpha-helices, beta-sheets, and loops! Thus, calculating the probability of it being a 2500 aa-long alpha-helix serves no purpose other than saying, oh well, it cannot be a 2500 aa-long alpha-helix. Well, it isn’t.

    How many more times should I repeat this?

    Oh, and by the way, the protein has homologs in archaea and bacteria.

    ETA: fixed some explanations, clarified a bit.

  10. Rumraket: You seem to have missed my point, which was that a presumably very unlikely event was nevertheless observed to occur naturally.

    But you didn’t answer my question. Why not assume man made it?

    And why not assume wind made stonehenge?

  11. Rumraket: Humans have built temples, sacrificial altars, arenas, observatories and ritualistic gathering sites, to the sun, moon and stars, to the seasons, and to animals, for tens of thousands of years.

    How do you know? BY looking?

  12. phoodoo: Why not assume man made it?

    Phoodoo we observed the iceberg to form this way naturally. Why then assume it didn’t? That makes no sense. Surely we can agree on this much?

    And why not assume wind made stonehenge?

    I believe I already explained why it is inferred to have been created by humans. It has the kinds of properties that are known to be the product of human design and manufacture, considering human psychology, culture and technology available at the time.

  13. phoodoo: How do you know? BY looking?

    Are you somehow trying to say that if we weren’t there to see it happen, no inference can be done? Is this Ken Ham’s “were you there” bullshit?

  14. colewd: Do I need a calculation to determine that you are the result of sexual reproduction?

    Do I need a calculation to determine gravity caused a pen to fall to the floor?

    I do need a calculation to determine if a “random” drop of coins caused 20 pennies on a table to come up heads.

    Identified deterministic process don’t need probability calculations to determine cause.

    Bill you are the one making the claim that A(plate tectonics creating the Mt Everest) is much more likely than B (a new protein evolving), and therefore that we can reject B while accepting A.

    That’s essentially what you’re saying here, that you reject one explanation because it is unlikely, and accept another because it is likely. I want you to back that up by doing the calculation.

    I think you know that if you were to do that, you would get very unlikely numbers in both cases and so your case for claiming that one is acceptable and the other is not, would be revealed as an example of the special pleading fallacy.

  15. colewd:
    Do I need a calculation to determine that you are the result of sexual reproduction?

    Nope.

    colewd:
    Do I need a calculation to determine gravity caused a pen to fall to the floor?

    Actually, yes. It could be that the speed with which it fell is much more than expected from gravitation alone, maybe some furious writer dropped it with a lot of strength, trying to break it.

    colewd:
    I do need a calculation to determine if a “random” drop of coins caused 20 pennies on a table to come up heads.

    This is misconceived. The random drop doesn’t “cause” the result. What “random” stands for “it just happened that way.” It’s not some “entity” or “force.” You do need a calculation to determine the expectation that you’d get all-heads from a “fair” coin drop.

    To get back to the 2500 amino-acids, what you’re doing is calculating the probability that the protein is all heads, when it’s a combination of heads and tails. What’s the probability of a combination of heads and tails Bill? Do you understand now why your math is misapplied?

    colewd:
    Identified deterministic process don’t need probability calculations to determine cause.

    Right. And evolution happens because of both deterministic and random events.

  16. Entropy,

    To get back to the 2500 amino-acids, what you’re doing is calculating the probability that the protein is all heads, when it’s a combination of heads and tails. What’s the probability of a combination of heads and tails Bill? Do you understand now why your math is misapplied?

    I am simply using Jock’s estimate as a conservative estimate as 2500AA’s in an alpha helix is as you say non functional but the simplest secondary fold. The problem is this sequence has to emerge with no selective advantage helping it along. I understand this is not a viable protein it is just an exercise in estimating secondary folds. I believe if we add a beta sheet the odds might get worse. I don’t see any random mechanism finding a viable secondary fold here.

    Do you understand the sequence that you are a claiming has homology to to PRPF8 does not have sequence homology?

    This family contains Prp8 domain IV, which adopts a RNase H like fold within its core structure but WITH LITTLE SEQUENCE SIMILARITY.

  17. Rumraket,

    Bill you are the one making the claim that A(plate tectonics creating the Mt Everest) is much more likely than B (a new protein evolving), and therefore that we can reject B while accepting A.

    I am making the claim that there is probably a deterministic mechanism we can model in the mountain case which does not exist in the biology case. One case we can make a scientific claim the other we cannot if a model is a minimum criteria. Some scientists claim a tested model is a minimum criteria. We cannot even model vertebrate conception to birth although there is hope here.

  18. To save colewd the trouble of locating my “31% of six-mers” estimate, here it is.
    If anyone does not understand the error that colewd is making, I will explain.
    It is ironic that colewd continues doing his naive multiplication on this thread when vjt’s sixth question is explicitly pointing out the error of P(A&B) = P(A) x P(B).

  19. DNA_Jock,

    It is ironic that colewd continues doing his naive multiplication on this thread when vjt’s sixth question is explicitly pointing out the error of P(A&B) = P(A) x P(B).

    The old the estimate is set up wrong canard. We have 140 orders of magnitude wiggle room here. The simpler equation works fine.

  20. Your math is sufficiently bad that 140 orders of magnitude is not a safe “margin for error”. Or “wiggle room”, as you put it.
    YOUR estimate for the probability of a 2500aa protein has just shifted from 10^-752 (i.e. 50%^2500 — remember that calculation?) to 31%^(2500/6) = 10^-212
    Your second estimate is still wrong fro multiple reasons, but given the fact that you were willing to shift it by 540 orders of magnitude without blinking an eye, what’s another 140 between friends?
    If you wish to use an approximation, the burden is on you to show that your approximation is fit for purpose.
    You cannot.

  21. colewd: I am making the claim that there is probably a deterministic mechanism we can model in the mountain case which does not exist in the biology case.

    What do you base that claim on? Now you that you have claimed it, what supports it? Can you predict the weather 300 million years into the future? Can you predict the convection currents in the mantle? At the resolution of individual atoms there is thermal noise and brownian motion taking place in the crust and mantle.

    One case we can make a scientific claim the other we cannot if a model is a minimum criteria. Some scientists claim a tested model is a minimum criteria. We cannot even model vertebrate conception to birth although there is hope here.

    Bill you’re blathering. It’s like you have this need to just say something, so you string words together and it’s irrelevant when it isn’t incoherent.

  22. colewd: The problem is this sequence has to emerge with no selective advantage helping it along.

    How do you know that?

  23. colewd: I don’t see any random mechanism finding a viable secondary fold here.

    You don’t see it, but fail to show why that is. This is the same thing over and over again.

  24. Rumraket,

    You don’t see it, but fail to show why that is. This is the same thing over and over again.

    There is simply too much wiggle room in the calculation vs evolutionary resources 10^50. If through probability estimates we can’t even get within 100 orders of magnitude of the available evolutionary resources of a secondary (I’ll reserve 40 orders for Jock’s contingency) fold for one protein of a major transition there is not much to argue about. The mechanism has to be deterministic.

  25. colewd: The problem is this sequence has to emerge with no selective advantage helping it along.

    Rumraket: How do you know that?

    colewd: What evidence do you have of a productive unfolded protein?

    So now that we have cleared up that you don’t in fact know that and now want to shift the burden of proof to me, despite it being you who made the claim, we can completely dismiss your mere assertion.

    And proceed here to an article documenting a case of that which you insist cannot happen, while also discussing other known examples in the introduction: Foldability of a Natural De Novo Evolved Protein.

    Not all proteins, however, require specific folding to function, and the relaxation of classic assumptions about structure/function relationships could make more radical mechanisms for molecular innovation far more plausible. Intrinsically disordered proteins (IDPs) that cannot fold independently are a sizable minority of proteins and serve a variety of biological functions (Dyson and Wright, 2005; Meszaros et al., 2007; Schlessinger et al., 2011; Tompa and Kovacs, 2010). IDPs vary widely in the level of disorder, from random coils to ‘‘pre-molten globules’’ to molten globule states, which are able to fold compactly and have high levels of secondary structure but lack specific tertiary structures (Dunker and Obradovic, 2001; Habchi et al., 2014; Uversky, 2002). Molten globule states may be functional (DeGrado, 1993; Pervushin et al., 2007; Vamvaca et al., 2004), and even the most IDPs can function through short linear binding motifs (Davey et al., 2012). A functional protein might therefore evolve de novo even if it cannot fold specifically. A protein born as an IDP might later evolve more native-like properties (Zhu et al., 2016) or continue indefinitely as a partially or completely disordered protein.

  26. colewd,

    I believe if we add a beta sheet the odds might get worse.

    You are determined to multiply simple probabilities. A vital component – a different kind of multipication – is routinely omitted. Not just by you, by nearly everyone on your side of things, at every level from organism down to protein segment. The fact that nucleic acids have the capacity of replication is vital. You can’t just ignore it; it is what drives life itself, at all scales. If the probability of 1 cell is X, the probability of N cells is not X^N, if that cell and its descendants.can replicate. Same goes for proteins (can we resist the ‘gotcha’ here? I know proteins don’t replicate directly).

    But at the genomic scale, there is an additional component: this replication is coupled with recombination – even in non sexual species. If you can duplicate a segment – which you can – it can be stuck virtually anywhere, and this changes the calculation by vast orders of magnitude. You don’t have to conjure up helixes, sheets and turns from random space. You can replicate one you already have and shove it somewhere else. Of course on insertion it has to align codon boundaries, which reduces chances by a factor of 3, but this isn’t going to trouble the probability gods unduly.

    This does not mean that proteins never come from previously untranslated DNA. But even there, genomic DNA sequences that will locally fold are over-represented, when compared to the truly ‘random’ sequence that would be squirted out by some kind of ‘Hoyle-o-matic’ sequence generator – a process for which I can see no biological analogue.

    It has to get going of course. But that is not a problem for evolution. It may be deemed impossible to get the system started, but once it’s started, probability calculations that ignore mechanism are worthless, and certainly no barrier to evolution.

  27. The propensity for arbitrarily picked intergenic regions to exhibit sequences with folding potential is an interesting question in it’s own right. The above cited example indicates that the intergenic region that gave rise to the Bsc4 protein actually coded for a protein that was capable of rudimentary folding already to begin with. (Interesting side note is that the Bsc4 protein isn’t essential, but apparently has a DNA repair function during nutrient poor conditions).

    That is however not a necessary precondition for de novo protein evolution from non coding DNA. Other examples are cited in that article where it is shown that a folding protein evolved from an intrinsically disordered ancestor that merely contained some functional motif.

  28. The other issue I’ll mention here is the bogus-ness of treating the problem as one of picking random acids from a 20-acid starting set. Or, in some ‘generous’ interpretations, allowing some kind of clustering on property, but then still using that as the basis of the computation – V acids in a string of length N is naively a space of V^N

    In fact there are about 500 possible amino acids, which some might imagine makes matters even worse, greatly inflating V. Does that cause my urbane smile to waver? Not a bit of it.

    Proteins are synthesised on the ribosome, and tRNAs are charged by enzymes, both giving much opportunity for stereoselective filtration of the wider set – only alpha acids (carboxyl and amino groups attached to the same carbon atom) and L acids (‘left-handed’ stereoisomers) can pass through this combined set of filters.

    Another constraint is biosynthesis. Organisms only deal with those acids that are available, generally synthesised either by themselves or by their food, making actuality a subset of possibility.

    And finally, we have an opposite tendency, increasing V: the capacity of a limited codon set to expand, by filling in STOPs.

    Now, what justifies the use of the resultant of these various processes and constraints – the 20 acid set of modern organisms – as a base for a probability calculation based on fully random picks? Nothing that I can see.

    Try this illustration. Imagine a primitive organism that has just three amino acids – let’s say hydrophobic, polar uncharged, charged. So the base V chosen for hokey probability calculations is 3. Now imagine this lineage gains a 4th assignment, then a 5th, and so on all the way up to 20. With each addition the space increases by an order of magnitude. Is it sensible to imagine that it thus becomes less easy to form proteins as this series progresses – as protein space gets bigger, by increasing V? You’ve actually added the possibility of subtle tuning, and increased the dimensionality of the space, which would increase the number of paths available, but no, apparently you’ve made evolution get progressively harder!

    The V^N brigade implicitly insist that the very act of adding acids to the set renders even the organism’s existing protein repertoire less likely…

  29. Allan Miller: But at the genomic scale, there is an additional component: this replication is coupled with recombination – even in non sexual species. If you can duplicate a segment – which you can – it can be stuck virtually anywhere, and this changes the calculation by vast orders of magnitude. You don’t have to conjure up helixes, sheets and turns from random space. You can replicate one you already have and shove it somewhere else.

    Allan Miller, prophet:

    A Molecular Portrait of De Novo Genes in Yeasts

    Abstract
    New genes, with novel protein functions, can evolve “from scratch” out of intergenic sequences. These de novo genes can integrate the cell’s genetic network and drive important phenotypic innovations. Therefore, identifying de novo genes and understanding how the transition from noncoding to coding occurs are key problems in evolutionary biology. However, identifying de novo genes is a difficult task, hampered by the presence of remote homologs, fast evolving sequences and erroneously annotated protein coding genes. To overcome these limitations, we developed a procedure that handles the usual pitfalls in de novo gene identification and predicted the emergence of 703 de novo gene candidates in 15 yeast species from 2 genera whose phylogeny spans at least 100 million years of evolution. We validated 85 candidates by proteomic data, providing new translation evidence for 25 of them through mass spectrometry experiments. We also unambiguously identified the mutations that enabled the transition from noncoding to coding for 30 Saccharomyces de novo genes. We established that de novo gene origination is a widespread phenomenon in yeasts, only a few being ultimately maintained by selection. We also found that de novo genes preferentially emerge next to divergent promoters in GC-rich intergenic regions where the probability of finding a fortuitous and transcribed ORF is the highest. Finally, we found a more than 3-fold enrichment of de novo genes at recombination hot spots, which are GC-rich and nucleosome-free regions, suggesting that meiotic recombination contributes to de novo gene emergence in yeasts.

  30. Rumraket,

    It would be very hard to rule out the possibility that segments of a de novo intergenic region had never been part of a functional protein – in a genome stuffed with gene, pseudogene, viral fragments and bits of broken transposon, and with several mechanisms of recombination, it would be hard to maintain a truly ‘ex-exon-sterile’ region over a long run. I’d imagine!

  31. Rumraket,

    No one living today witnessed the making of stonehenge or the iceberg. So the only thing you are basing the conclusions on are that it looks like something nature can make or it looks like something man could make.

    So if this is a good enough criteria for deciding if something is designed or not, then materialists conclusion that life wasn’t designed is not very convincing. Nature can make squares but not pyramids, and bacteria but not skyscrapers or tide detergent.

    Its all just what one feels like speculating.

  32. colewd:
    I am simply using Jock’s estimate as a conservative estimate as 2500AA’s in an alpha helix is as you say non functional but the simplest secondary fold.

    I know that’s what you’re doing, you said so from the very beginning. You keep missing my point: calculating a probability that it will be a 2500 aa-long alpha-helix doesn’t help your case. It only highlights your problems: you’re using the wrong math, the wrong probabilities, and the wrong model to try and justify your improbability assertions. What good does it make to your case if you insist on missing the point? Do you really think that an unbiased reader would buy into your calculation when I’m showing them that expecting several kinds of structures in a protein better matches what the actual protein looks like?

    colewd:
    The problem is this sequence has to emerge with no selective advantage helping it along.

    False. I truly don’t know what makes you think this. Form and function come together.

    colewd:
    I understand this is not a viable protein it is just an exercise in estimating secondary folds.

    I know this is what you think it is. But it’s an exercise of demonstrating how you fail to properly considering the scenarios before applying your math.

    colewd:
    I believe if we add a beta sheet the odds might get worse.I don’t see any random mechanism finding a viable secondary fold here.

    If the sequence doesn’t have a propensity to produce an alpha-helix it might have a propensity to form a beta-strand, or a loop. That adds up to the probability of finding some structure. Like the coins, it’s either tails or heads. What’s the probability that you’ll have either heads or tails Bill? It’s not 0.5 * 0.5 = 0.25, it’s 0.5 + 0.5 = 1. So, in multiple drops, what’s the probability of having a combination of heads and tails? Given that, we expect a 2500 aa-long protein to have stretches of alpha-helix, beta-strands, loops.

    Starting to get it now?

    colewd:
    Do you understand the sequence that you are a claiming has homology to to PRPF8 does not have sequence homology?

    Even though they’re related, homology and similarity are not synonyms. Proteins can be homologous, but have their histories separated by such a long time span that they are no longer detectable as homologs by their similarity. In this case, however, though the similarity might be “low,” both the statistics and the structure are enough to establish this protein as homologous to RNA-acting proteins in bacteria and archaea.

  33. Allan Miller,

    It has to get going of course. But that is not a problem for evolution. It may be deemed impossible to get the system started, but once it’s started, probability calculations that ignore mechanism are worthless, and certainly no barrier to evolution.

    It’s a huge problem for evolution. The observation of the sequence falsifies the theory because the probability problem is so large for the claimed mechanism. We can make continent estimates to cover the details. All your side can do is say it cannot be calculated because this mechanism may make a difference. That claim is not scientific.

  34. Entropy,

    If the sequence doesn’t have a propensity to produce an alpha-helix it might have a propensity to form a beta-strand, or a loop. That adds up to the probability of finding some structure. Like the coins, it’s either tails or heads. What’s the probability that you’ll have either heads or tails Bill? It’s not 0.5 * 0.5 = 0.25, it’s 0.5 + 0.5 = 1. So, in multiple drops, what’s the probability of having a combination of heads and tails? Given that, we expect a 2500 aa-long protein to have stretches of alpha-helix, beta-strands, loops.

    Ok, you have made a counter argument. How if all does this improve the probability? Lets take Weasel and give it 2 targets. Do you think it will finish faster? Will it finish at all? I have given Jock 40 orders of magnitude wiggle room for contingencies like the one you brought here. I don’t think you can reduce the improbability by 40 orders of magnitude with your claim but please go for it.

    I also give Rum lots of credit for bringing disordered proteins into the discussion.

    Lets bring back some perspective to this based on Axe’s claim that the probability problem is so large it is obvious. We are talking about the first observed fold of one protein of a 200 protein complex that is not the most complex system in a multicellular eukaryotic cell.

  35. Entropy,

    Even though they’re related, homology and similarity are not synonyms. Proteins can be homologous, but have their histories separated by such a long time span that they are no longer detectable as homologs by their similarity. In this case, however, though the similarity might be “low,” both the statistics and the structure are enough to establish this protein as homologous to RNA-acting proteins in bacteria and archaea.

    The dissimilar sequence is evidence against it evolving. The truth is that even if the segment had been identical to the segment in PRPF8 it would do very little to help your argument.

  36. From the abstract

    Therefore, identifying de novo genes and understanding how the transition from noncoding to coding occurs are key problems in evolutionary biology.

    Showing they evolved at all is the problem for evolutionary biology.

  37. Allan Miller,

    The V^N brigade implicitly insist that the very act of adding acids to the set renders even the organism’s existing protein repertoire less likely…

    Alan this is the equation. You can modify the result based on substitutability as Axe and others have but the equation for estimating a sequence remains valid.
    If I were playing the game like the NCSE I would call you guys math deniers 🙂

  38. colewd,

    You keep missing it Bill. Check this carefully:

    It’s not 0.5 * 0.5 = 0.25, it’s 0.5 + 0.5 = 1. So, in multiple drops, what’s the probability of having a combination of heads and tails?

    See? If you consider that amino–acid sequences will have varying tendencies you’ll end up concluding that a 2500 aa-long protein will be a combination of alpha-helices, loops, turns, beta-sheets, etc. Guess what it’s like? Yeah! A mixture of loops, alpha-helices, loops, turns.

    You’re calculating the probability that it would be an alpha-helix, because if you considered that it’s a combination of things you’d produce a large probability, and you would not be happy about it, since it contradicts your stance.

    With the coins it’s easier to understand. Let’s say I had some combination of heads and tails. You then tell me, no, this is impossible because the probability of heads is 0.5, just for one coin. The probability for 2500 heads is so small it won’t happen! But I have a combination of heads and tails, what’s the probability that I’d have a combination of heads and tails Bill? Yet you go back to all heads, but it’s not all heads! So, when will we reach some understanding Bill?

  39. colewd:
    The dissimilar sequence is evidence against it evolving.

    That’s an unquantified dissimilarity, and, again, it doesn’t take away the stats, the structure and the function, Bill. Come on.

    colewd:
    The truth is that even if the segment had been identical to the segment in PRPF8 it would do very little to help your argument.

    It’s not an argument, it’s a scientific result, and it wasn’t me who analyzed those proteins and found them to belong into the same domain family (aka homologs).

  40. colewd: It’s a huge problem for evolution.

    It clearly isn’t.

    The observation of the sequence falsifies the theory because the probability problem is so large for the claimed mechanism.

    Why?

    We can make continent estimates to cover the details.

    What does that mean? What is a “continent estimate” and to cover the details of what?

    All your side can do is say it cannot be calculated because this mechanism may make a difference.

    What are you even talking about? You’re blathering again Bill.

  41. colewd: Lets bring back some perspective to this based on Axe’s claim that the probability problem is so large it is obvious.

    What probability problem? You never get around to showing that there is a probability problem. All you ever do is insist that because the space of sequence length is large, that means functions can’t evolve. You never actually show why. You just declare it.

  42. colewd: The dissimilar sequence is evidence against it evolving.

    That makes zero logical sense Bill. If they diverged from a common ancestor, they would be expected to become increasingly different with time.

    Dissimilarities in sequences cannot constitute evidence against evolution when they are expected on evolution.

    Bill please just try to think for a moment. Stop the arguing mindset, get out of the “I have to show that they are wrong” mindset, and get into the “I have to figure out what is true” mindset instead. When you are in the “I have to show they are wrong” mindset you can’t think straight.

    Seriously, look at what you just said. Sequence dissimilarities are evidence AGAINST evolution? THATS NUTS!

    Please, for the love of all that is good, get out of that debilitating mindset just for ONCE in your goddamn life.

    The truth is that even if the segment had been identical to the segment in PRPF8 it would do very little to help your argument.

  43. Entropy,

    With the coins it’s easier to understand. Let’s say I had some combination of heads and tails. You then tell me, no, this is impossible because the probability of heads is 0.5, just for one coin. The probability for 2500 heads is so small it won’t happen! But I have a combination of heads and tails, what’s the probability that I’d have a combination of heads and tails Bill? Yet you go back to all heads, but it’s not all heads! So, when will we reach some understanding Bill?

    Your analogy fails because your are not dealing with heads and tails. You are dealing with 20 amino acids. You have a combinatorial problem and there is no way out of this. The only way to support your claim is if all 20 amino acids were equally substitutable and even Allan and Jock won’t support you here.

  44. Rumraket,

    Seriously, look at what you just said. Sequence dissimilarities are evidence AGAINST evolution? THATS NUTS!<

    Strong argument Rum 🙂

  45. Rumraket,

    What probability problem? You never get around to showing that there is a probability problem. All you ever do is insist that because the space of sequence length is large, that means functions can’t evolve. You never actually show why. You just declare it.

    Sure I do Rum. Please re read the posts. You are facing long sequences with limited substitutability. You need to show almost complete substitutability for evolution to have a prayer. Thats the claim entropy is trying to make but the empirical evidence shows this is false as gpuccio has been trying showing you guys. I think its time for you to start to face the reality that the theory that life’s diversity can be explained by reproduction alone is toast.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.