More Recent Comments

Tuesday, October 05, 2010

The Edge of Evolution

There are two main problems with Intelligent Design Creationism. The first is that the IDiots never have anything positive to offer by way of explanation. They complain about how evolution can never do this or that but they never give us a better explanation based on their beliefs. The second problem is that the IDiots often get their science wrong when they complain about evolution. Sometimes this is deliberate, but in many cases it's because they just don't understand what they're criticizing.

Their criticisms of evolution are based on the notion of a false dichotomy. They think that there are only two choices: their conception of evolution, or Intelligent Design Creationism. Thus, if they can refute their version of evolution it follows that creationism must be true.

People often make the claim that Intelligent Design Creationism isn't science. That may be true if you only think of it as promoting the idea of an intelligent designer but even there it depends on how you define science. However, much of the Intelligent Design Creationism literature isn't about defending creationism, it's about attacking evolution and those arguments definitely fall within the definition of science. As scientists, we have to deal with all the objections to evolution no matter what the motives of the challengers.

I think it's somewhat simplistic to dismiss all the Intelligent Design Creationist literature on the grounds that it's not science. Some of it has all the earmarks of science, it's just bad science. And bad science isn't limited to IDiots. I think there are many Theistic Evolutionists who are also guilty of promoting bad science and there are many atheist scientists who are just as guilty. The peer-reviewed scientific literature is full of examples.

Although it makes my American friends cringe, I favor teaching the controversy. It's the only way to show students the difference between good science and bad science.

Some of the Intelligent Design Creationists can craft pretty convincing arguments against evolution. It takes a lot of work to refute them. I going to give you an example of such an argument from The Edge of Evolution by Michael Behe. Let's see how you do.



The Two Binding Sites Rule

Behe's version of the history of life requires a God who intervenes quite frequently to create specific mutations that are almost impossible to account for by random mutation. Behe makes a good case for the problems with random mutation. In fact, his arguments are similar to those put forth by the mutationist camp—a group that I'm in sympathy with. Most biologists would not be able to refute Behe's arguments because they would agree with some of his false premises.

Behe's "Two Binding Sites Rule" is a good example. He argues that in order for two proteins to interact, evolution needs to create a small patch on the surface of each protein where five or six amino acid side chains become compatible with binding. Some of these changes could be neutral so they could arise independently but the analysis of hundreds of known binding sites shows that many of the mutations would be detrimental if they occurred by themselves—a single charged amino acid residue on the surface, for example.

It looks like you need to wait for three or four specific mutations to occur simultaneously in order to get a moderate interaction between two proteins that did not originally bind to each other. And these can't be just any proteins, they have to be proteins where there is a selective advantage to forming a complex. The example I've chosen is a bacterial photosynthesis reaction center where four polypeptides (gold, blue, green, purple) interact with each other and with multiple cofactors (space-filling molecules) to form a very complicated structure. Presumably, there was a time in the past when some of these proteins didn't bind to each other or to the cofactors. Over time, evolution favored variants that could form the complex. How could this happen according to evolutionary theory?

Studies on in vitro mutagenesis show that the probability of forming any de novo binding site is very low. For example, it's quite difficult to engineer specific antibodies that will bind to a particular antigen. The data shows us that you need a library of more than one billion antibody molecules in order to get one that will bind. Those one billion mutations are far from random. They are engineered so that they are confined to a small patch on the surface of the antibody where it is known that other proteins can potentially bind.

Behe argues from this evidence that the probability of creating a new binding site by random mutations is exceedingly small. So small, in fact, that such mutations would only arise in very large populations after several hundred million years of evolution. He bases his argument on some experiments he describes in the first few chapters or his book.

Behe points out that it is sometimes very difficult for the malaria-causing parasite, Plasmodium falciparum, to develop resistance to some drugs used to treat malaria. That's because the resistance gene has to acquire two specific mutations in order to become resistant. A single mutation does not confer resistance and, in many cases, the single mutation is actually detrimental. P. falciparum can become resistant because the population of these single-cell organisms is huge and they reproduce rapidly. Thus, even though the probability of a double mutation is low it will still happen.

If the probability of a single mutation is about 10-10 per generation then the probability of a double mutation is 10-20. He refers to this kind of double mutation as CCC, for "chloroquine-complexity cluster," named after mutation to chloroquine resistance in P. falciparum.1 Behe's calculation is correct. If two simultaneous are required then the probability will, indeed, be close to 1 in 1020.

Let's see how this relates to the evolution of protein-protein interactions. Here's how Behe describes it on page 135 of his book.
Now suppose that, in order to acquire some new, useful property, not just one but two new protein-binding sites had to develop. A CCC requires, on average, 1020, a hundred billion billion, organisms—more than the number of mammals that has ever existed on earth. So if other things were equal, the likelihood of getting two new binding sites would be what we called in Chapter 3 a "double CCC"—the square of a CCC, or one in ten to the fortieth power. Since that's more cells than likely have ever existed on earth, such an event would not be expected to have happened by Darwinian processes in the history of the world. Admittedly, statistics are all about averages, so some freak event like this might happen—it's not ruled out by the force of logic. But it's not biologically reasonable to expect it, or less likely events that occurred in the common descent of life on earth. In short, complexes of just three or more different proteins are beyond the edge of evolution. They are lost in shape space.
We're all pretty knowledgeable here but how many of you can immediately refute that argument? If you can't then you have no business accusing Behe of being stupid or silly and of dismissing his book as just another example of creationist ignorance. The correct explanation of the problem will undoubtedly appear soon in the comments. Before peeking, why not try and see how you would answer Behe if you were debating him in front of a large audience of creationists?


1. Behe may have been wrong about the specific chloroquine resistance mutation he used as an example. The two mutations may not have occurred simultaneously. Nevertheless, the principle is sound. If the single mutations are detrimental then you need both to get resistance and the probablity of two such mutations occurring together is 10-20.

64 comments :

Pigasus said...

Doesn't the Lenski experiment crush this argument like a grape?

K said...

I just recently mentioned this very thing (that if I were in a debate with Behe I would get schooled) in a discussion about debating.

The problem I see is the Texas sharpshooter fallacy.

lee_merrill said...

Yay--I'm all ears. And a little mouth: Behe quotes the Lenski experiment, may it be said. And he's talking about general mutations, not one specific result or binding site, so it seems he's not marking the target after the shot.

c-serpent said...

Um.... I'm not sure but I'll give it a try so I can learn something.

He assumes that to acquire ANY useful property and thus for the protein to stick around at all there must x number of simultaneous mutations at the particular binding site in his example. While his hundreds of binding sites analysis no doubt showed that most of these mutations were detrimental to the current function involving that particular binding site this ignores any previous functions that the protein may have had, probably entirely unrelated to the location of that binding site. Mutations could accumulate in regions of the protein distant from the functional areas that would eventually result in the formation of the required sequence for the binding sites.

Isn't the literature replete with examples of proteins with one function co-opted for other purposes? His favorite example the bacterial flagellum from his first book comes to mind.

Harriet said...

Why do these mutations have to happen at the same time? One can happen (and possibly be useful for some other reason), get fixed by natural selection, then the second one can happen.

In probability theory, this is the difference outlined here.

Anonymous said...

An English graduate here, just looking at the way the question's framed.

Is the problem the word 'simultaneous'?

If so, wouldn't that make this just another irreducible complexity argument?

Ooh, am I right? If so, do I win something?

Anonymous said...

Yeah, do I win something?

Sparky said...

Well, I see two problems. The first is the supposition that neutral or even detrimental mutations will not be propagated or become fixed in the population. Our ordinary experience of the world tells us that this is not so, as deleterious mutations and congenital defects of various minor kinds are present in almost every individual we ever meet. Unless it is immediately fatal any mutation may propagate through a population given the right conditions. So, the requirement that the mutations be simultaneous is bogus. A "harmful" mutation can very likely be propagated long enough for a compensating mutation to occur, particularly under conditions where no set of genes is robustly reproducing.

However, I don't think we even need to rely on the proven power of living organisms to tolerate deleterious mutations. If someone tells you to turn TAP into RUG by changing only one letter this is of course impossible. But it is trivial to change these words into each other with a series of one-letter changes (TAP TAG RAG RUG) and even keep a meaningful word at every step along the whole path. Similarly, an interaction of hydrophobes (say F-F) can be turned into a salt bridge (D-R) by a short chain of single changes (F-F F-H D-H D-R) in which each intermediate step retains the property of having a favorable interaction energy. There is no particular reason to insist that the only mutations made going from AB to CD be A to C or B to D.

DK said...

Obviously, the 10^(-20) number is wrong. Nobody *designed* the interaction interface. In other words, organisms don't just sits there and wait for the right mutations to occur to create this particular interface ("or else", suggests Behe).

In reality, this is how protein interfaces come to existence:
1. Thousands of polypeptides in the cell.
2. Each of them can have hundreds of near neutral mutations.
3. All of these random mutations in all of the proteins create a *huge* number of new protein-protein interfaces.
4. Whenever one of these newly created interfaces happens to be beneficial in any way, it gets a chance to be fixed by selection.

The key difference: Behe's picture in effect assumes design to begin with (in order to prove it) while the perfectly random evolutionary picture merely says that anything goes and out of almost infinite number of combinations some get realized by chance (*). Obviously, the latter picture does not require any two things to have happened simultaneously.

(*) Larry, this is the primary reason why independent "inventions" of anything complex are so unlikely. As in protein fold - which is a pretty complex thing. Thus, convergent evolution of whole protein folds has probably never happened. :-)

c0nc0rdance said...

There are three things I see wrong with Behe's combinatorial argument:

1. It assumes that only a single sequence will be functional. He imagines a single fitness peak in a sea of non-functional changes. Axe makes the same mistake. Proteins may be found to weakly interact with a very high percentage of possible sequences. These weak interactions create a fitness slope that selection can walk up. ID Creationists spend so much time characterizing the cell as a series of little machines, they forget cells are squishy, sticky balls of fat, protein, and sugar.

2. There is an assumption of a static environment, or a single selection. As though the cell were a uni-tasker, only working on one optimization at a time. In truth, every protein has a chance of being duplicated, diverging, and filling a niche in the cell. All these modular protein domains, constantly being shuffled and recombined, activated and transposed. It's only this over-simplified point mutation and natural selection model that produces such silly numbers.

3. When we compare sequences between organisms, we find thousands of different sequences for the same protein. Even histones show considerable diversity, yet they all perform the same function. How can there be only one possible sequence if there are over a hundred functional sequences in Genbank?

I attended the SMU Event, and made the recording seen in an earlier post. I put these issues to Doug Axe, and was astounded at his reply. If you are interested:
http://www.youtube.com/watch?v=0MHAO4JE0G0

I've also got proof that Stephen C Meyer is a witch:
http://www.youtube.com/watch?v=t4l0T31ovP0

Gyalogtank said...

Many, many errors with this argument.

A.) Behe never defined what Kd does he consider as "binding". Two proteins can form complexes by very strong or very veak interactions, weak interactions are more common, as many more proteins can form a weak complex with each other, than a strong one. He deliberately cites an experiment, where the goal was to create proteins which form a complex stable enough to maintain the comples at 80 C, which interaction is much much stronger than the majority of protein complexes found in a human body.

Behe never states, at what Kd does he assume the two prtoeins "interact", and never ever examines, how stable a protein complex should be to confer any advantage. Weak interactions could work at, let's say 1% of the activity of the activity observed today, and still be useful for the organism. It's reall just some random numbers multiplied to get a sufficiently high number, to make evolution look "impossible".

B.) He assumes, that every mutation has to arise at precisely the same moment. In his book, he cites two examples cloroquine and S/P resistnace in Plasmodium falciarum. He states, that cloroquine resistance is a CCC because two point mutations are required for this trait to arise. HE also states, that S/P resistance is a CCC, but in this case, three point mutations are needed, and he does not adress this interesting phenomenon, namely, that his second example is more or less out of the "edge of evolution", it is highly imporpable, or would take millions of years to arise.

He also fails to address a simple question: Is a CCC is the edge of evolution, then why is malaria still a threat? Treat malaria with cloroquine, and any cloroquine resistant strains should be treated with S/P cocktail. Or treat every malaria infection with cloroquine and S/P, as a single CCC is the edge of evolution, and by according to Behe's calculations every mutation has to occour at the very same moment, it is impossible, that any Plasmodium falciparum could be resistant to both drugs, as that would take two CCCs, which is impossible according to Behe. But these double resistant strains can be found in nature. How did they arise? These cells clearly harbour two CCCs. BUt this is impossible, according to Behe.

Susannah Anderson said...

Is this too simplistic? I am not a scientist, and it's a half-century since I was in school, but my immediate thought was that he is assuming that all 10 to the fortieth mutations have to be "in" before one trio is selected, which is ridiculous. The three could, hypothetically, appear in the first generation.

Or am I missing something?

lee_merrill said...

Two quick points, Behe's argument I believe only requires statistically independent mutations, so the probability of two such mutations occurring simultaneously is the same as the probability of them occurring at different times. And the numbers are based on what evolution actually does, so mutations being fixed by drift, etc. is not being overlooked.

Anonymous said...

One quick point - Lee Merrill hero worships Behe and will never be convinced that he is wrong.

Divalent said...

One problem (which I think was raised by several others above; I'll just try to state it a bit more starkly) is that he assumes that this one pair of mutations are the:

1) only possible way to have
2) those two proteins interact in a
3) functionally beneficial way to have the organism have
4) the specific function under consideration.

It's like saying the only possible way Larry could have arrived at work today was to do each and every action action precisely as he did in exactly the same order at precisely the time he did them from the moment he got up at exactly the time he actually got up.

Ghostrider said...

Lee Merrill has had a man-crush on Behe for years. Behe's argument in EoE has been beaten into a fine pink mist by dozens of people over at TWeb where Merrill hangs out, but Merrill just keeps on blindly regurgitating Behe's claims. It's really sad.

Steve LaBonne said...

The two mutations may not have occurred simultaneously. Nevertheless, the principle is sound. If the single mutations are detrimental then you need both to get resistance and the probablity of two such mutations occurring together is 10-20.

I don't think this is correct if the mutations are NOT simultaneous. Only if they are simultaneous is it legitimate to multiply the probabilities of the two mutations to get the probability of the double mutation.

Think of a dice-rolling example. What's the probability of throwing boxcars? 1 in 36. But what's the probability of rolling one die multiple times and having it at some point come up six, followed by rolling the second die and having it at some point also come up 6? THAT depends completely on the number of rolls you're allowed with each die. It will approach 1 at the limit as the number of allowed rolls increases.

Simultaneity is crucial to Behe's argument and is his major error, as far as I can see. You can get even the single-mutation frequency of 10^10 ONLY by requiring that it happen in a PARTICULAR replication event, so that the total probability comes out equal to the per-generation probability And this constraint in turn comes entirely from the simultaneity constraint because only that can force the mutation to occur in a particular generation(i.e. the same one in which the second mutation also occurs). I welcome correction if I've missed something.

Anonymous said...

"They think that there are only two choices: their conception of evolution, or Intelligent Design Creationism."

I've got a simple rule. Those who use the term "Intelligent Design Creationism" lose. And only one side is using the term.

Steve LaBonne said...

I've got a simple rule.

And we'd be happy to tell you where you can put it.

Anonymous said...

"two such mutations occurring simultaneously is the same as the probability of them occurring at different times"

This is just the bombardier beetle argument.

You're starting from a false premise of design. That the two chemicals in the bombardier beetle / two sticky patches on a protein have no function at all except to be together, that they have to happen at *exactly* the same time, and that they came together to fulfill some purpose.

MattK said...

Steve,

The specific cases referred to here are those in which the either of the two (or more) mutations that confer an advantage together are deleterious by themselves. So if one shows up without the other, it will be quickly eliminated by selection unless it is only very mildly deleterious. This effectively limits the 'number of rolls' to a small number - after that the mutant is eliminated from the population.

Steve LaBonne said...

MattK, where are the data that all of the mutations that can combine to provide chloroquine resistance are 1) detrimental at all, 2) strongly detrimental? Even Larry only says that "in many cases" ONE of the mutations "is actually detrimental".

Allowing multiple rolls of only ONE of the dice will still make Behe's number way off. This doesn't help him much unless BOTH mutations are SUFFICIENTLY detrimental to be subject to strong purifying selection. Only that constraint would be sufficient to require that they be (nearly) simultaneous.

lee_merrill said...

> Thornton: Behe's argument in EoE has been beaten into a fine pink mist by dozens of people over at TWeb where Merrill hangs out,

So then what is your answer to Prof. Moran's question?

If Behe's arguments have been demolished as you say, there should be no problem in giving your answer here.

Steve LaBonne said...

By the way, this is all academic anyway, because the premise which Larry accepts for the sake of argument does not at all (as is typical with Behe) correspond to what's actually in the literature:
http://tinyurl.com/2b5rlhu

Anonymous said...

If there are only 2 MDR phenotypes: dually susceptible and dually resistant, and only one genome with dually resistant genotype, then I would guess that it would be very hard for evolution to find this genotype. However, I would guess that there is distribution of resistances levels over genome space. And likewise selection for otherwise deleterious alleles changes the fitness landscape, opening space to place compensating mutations. Its a long walk to the Northwest Territories, and you will probably need a nicer jacket not magic.

Also I would imagine that neutral diffusion over the common genome space of the sufficiency good phenotypes creates a kind of potentiation where phenotypic change can happen quite rapidly. That is, the fitness potential of each allele changes in a dramatically new environment. A previously 'neutral' mutation may in fact have a nice compensating effect for a partially resistant genotype.

Larry Moran said...

Steve LaBonne says,

By the way, this is all academic anyway, because the premise which Larry accepts for the sake of argument does not at all (as is typical with Behe) correspond to what's actually in the literature:
http://tinyurl.com/2b5rlhu


Behe did his best to support his claim by referencing the scientific literature. We now know that the specific mutation he referred to may not have required two simultaneous mutations.

That doesn't refute the main point which is that the probability of two simultaneous mutations is, indeed, 10^-20.

Keep in mind that Behe's faults are no more egregious than those of many scientists who write popular books. That doesn't mean he's a *good* scientist. What it does mean is that his arguments are scientific, not religious.

Larry Moran said...

Steve LaBonne says,

Allowing multiple rolls of only ONE of the dice will still make Behe's number way off. This doesn't help him much unless BOTH mutations are SUFFICIENTLY detrimental to be subject to strong purifying selection. Only that constraint would be sufficient to require that they be (nearly) simultaneous.

I think you're close to understanding the main problem with Behe's argument.

He assumes that deleterious mutations will always be rapidly eliminated from the population. That's consistent with the common understanding of evolution so it appears to set up an insoluble problem.

However, you and I (and many others) know that evolution doesn't work that way. There's a lot of sloppiness and accident so it's quite possible for inefficient proteins to hang around for a long time. You could get the same effect by gene duplication and messing with the spare copy.

Steve LaBonne said...

I'm not sure I'd characterize lying about the literature, and positing (contra said literature) a clearly false mechanism of simultaneous mutations for chloroquine resistance, as a "scientific argument"- not even a bad one. Lying for Jesus, more like.

Larry Moran said...

Divalent says,

It's like saying the only possible way Larry could have arrived at work today was to do each and every action action precisely as he did in exactly the same order at precisely the time he did them from the moment he got up at exactly the time he actually got up.

Yes. That's another valid scientific criticism of Behe. There may be millions of possible mutations that would lead to the same result.

Larry Moran said...

Steve LaBonne says,

I'm not sure I'd characterize lying about the literature, and positing (contra said literature) a clearly false mechanism of simultaneous mutations for chloroquine resistance, as a "scientific argument"- not even a bad one. Lying for Jesus, more like.

That's unfair. I'm pretty sure that Behe isn't lying.

What do you think of someone like John Mattick and his arguments against junk DNA? Those arguments are just as bad as Behe's but we don't accuse him of lying for Jesus, do we?

I can think of other scientists who are worse than Behe. Do you remember Jefferey Schwartz's book "Sudden Origins"?

Gyalogtank said...

"However, you and I (and many others) know that evolution doesn't work that way. There's a lot of sloppiness and accident so it's quite possible for inefficient proteins to hang around for a long time."

That's why Behe's argument is especially tricky. The Plasmodium falciparium is not a good model system for metazoans, as it spends most of its life cycle as haploid cells. Haploid cells have only one allele of each of their genes, so every recessive mutation manifests in the phenotype. A diploid genome can harbour a lot of recessive lethal mutations, without experiencing any problems. A haploid organism simply evolves slover.

The malaria parasite divides asexually most of the time, so the majority of the "new generations" with which Behe wants to use as a model organism of all life, are simply the clones of a single cell. Sexual reproduction has a definite advantage, alleles are re-mixed in every offspring, but te Plasmodium falciparum is a poor model for evolution.

Behe also considers every P. falciparum generation to be equal to a human eneration, hovever in longer living organisms, simply more mutations occour between generations. Counting the mutation rate of P. falciparum and H. sapiens sapiens, it is more or less the same when expressed in mutations/year, however a human generation is ~20 years, and a malaria generation can be measured in days.

Larry Moran said...

Gyalogtank says,

Behe never defined what Kd does he consider as "binding". Two proteins can form complexes by very strong or very veak interactions, weak interactions are more common, as many more proteins can form a weak complex with each other, than a strong one. He deliberately cites an experiment, where the goal was to create proteins which form a complex stable enough to maintain the comples at 80 C, which interaction is much much stronger than the majority of protein complexes found in a human body.

Behe never states, at what Kd does he assume the two prtoeins "interact", and never ever examines, how stable a protein complex should be to confer any advantage. Weak interactions could work at, let's say 1% of the activity of the activity observed today, and still be useful for the organism. It's reall just some random numbers multiplied to get a sufficiently high number, to make evolution look "impossible".


Good point. It's one of the important problems with Behe's argument but in order to appreciate it you have view the workings of a cell as being somewhat sloppy and not fine tuned.

Most people don't hold that view so they miss the point you just made.

I would add that there are huge numbers of proteins that interact weakly with one another even though there's absolutely no reason for them to do so. There are even examples of fairly strong interactions that are not biologically relevant. They pop up all the time as false positives in interaction networks.

lee_merrill said...

> Prof. Moran: There's a lot of sloppiness and accident so it's quite possible for inefficient proteins to hang around for a long time. ... There may be millions of possible mutations that would lead to the same result.

Would this be redefining the edge, though? Certainly if you start with selectable weak interactions, and make them stronger with new mutations, or if there are many millions of mutations that produce a new interaction, then you can get there quickly. But that would be a different edge. And Behe points out that "Cellular proteins almost always are made with just one sequence" (p. 133).

But maybe "millions of different paths" was meant, if so, the statistical independence of the mutations means the path doesn't matter.

> Gyalogtank: It's really just some random numbers multiplied to get a sufficiently high number, to make evolution look "impossible".

Well, the question is if the edge Behe marks here is correct. And Behe acknowledges that evolution can produce new structures--just not structures that require more than four or more new independent mutations.

And is this analysis correct? "One way to get a new binding site would be to change just five or six amino acids in a coherent patch in the right way. This very rough estimation fits nicely with studies that have been done on protein structure. ... But five or six amino acid substitutions means that reaching the goal requires five or six coherent mutational steps..." (p. 134).

Then we have "the great majority of proteins in the cell work in complexes of six or more. Far beyond that edge" (p. 135).

If either of these statements is correct, then it would seem the edge is practical.

lee_merrill said...
This comment has been removed by the author.
lee_merrill said...

> Lee: ... two such mutations occurring simultaneously is the same as the probability of them occurring at different times.

> Anonymous: This is just the bombardier beetle argument.

Actually, it's a mathematical point, it follows from statistical independence.

Chris said...

It sounds to me as though the mutations being referred to in the original argument are point mutations that only change one specific amino acid at a time. However, there are mechanisms that can alter several amino acids in a protein sequence in a single event, such as indels (insertions/deletions) and frame shift mutations.

Siabolourani said...

Ok. So the probability of this specific event (CCC) happening is 10^{-20}. But this specific event has happened. There are other specific events that didn't happen. Let's call one of these events BBB. The probability of BBB happening is also 10^{-20} (assuming that this event is also a double mutation). This BBB might have done something else to malaria-causing parasite (cause them to go crazy and kill other malaria-causing parasites). Then Behe would be arguing that the probability of BBB happening is extremely low and therefore it is unlikely that it happened by chance.
I give you another example. The probability that the roulette wheel gets the sequence 12, 10, 30, 0, 3, 00, 4 one after another in a casino in Montreal, is very low. But after the wheel gets this sequence, we can't argue that this probability was low and the fact that it happened means that there is a hand of God tweaking the roulette table.
But if someone told you this sequence prior to the wheel turning and it happened, then it is almost improbable that he just got lucky. He probably has a hand in it.

Anonymous said...

I give you another example. The probability that the roulette wheel gets the sequence 12, 10, 30, 0, 3, 00, 4 one after another in a casino in Montreal, is very low. But after the wheel gets this sequence, we can't argue that this probability was low and the fact that it happened means that there is a hand of God tweaking the roulette table.

Yeah, this was covered in the second reply by Kel. It's the sharpshooter fallacy.

Steve LaBonne said...

Larry, did John Mattick ever pretend to do a thorough literature search on a topic he was writing a book about, and then pretend he didn't find anything when in fact there were a lot of papers in the area? Did John Mattick ever get on a witness stand and swear there were no papers in an area, whereupon the attorney questioning them was able to plop a big stack of them in front of his nose? If the answers to those questions are "yes" then I'll concede that Behe is no more dishonest than Mattick. Otherwise...

Sorry, Behe lies for money. It's how he makes his living nowadays. I've never been convinced that he believes his own BS.

Anonymous said...

"Actually, it's a mathematical point"

No, it's not.

They don't evolve simultaneously *to* stick together. There wasn't a plan with this endgame in mind.

They don't need to evolve simultaneously, they just have to be around at the time.

Sharks eat people. They have mouths that fit neatly around humans, teeth that can bite through them. Strident new atheists who worship Darwin in basement covens insist, against the word of the Lord, that sharks evolved 420 million years ago. They have a silly myth that human also evolved, but can't even agree when - but let's say 200,000 years ago.

But if that was the case, then surely all the sharks would have starved to death hundreds of millions of years ago?

... and that's what this protein argument is, it's just talking about things that Spielberg didn't make a movie about, so it's harder to picture.

Anyway if evolution was true, why wouldn't sharks have evolved into people by now, because that's the ultimate purpose of evolution? They've had hundreds of millions of years, and haven't even evolved legs!

For more explanation about sharks:

http://www.answersingenesis.org/creation/v23/i2/sharks.asp

Gyalogtank said...

lee_merill:

> Lee: ... two such mutations occurring simultaneously is the same as the probability of them occurring at different times.

> Anonymous: This is just the bombardier beetle argument.

"Actually, it's a mathematical point, it follows from statistical independence. "

"Well, the question is if the edge Behe marks here is correct. And Behe acknowledges that evolution can produce new structures--just not structures that require more than four or more new independent mutations."

You surely have read the second half of my comment. If you are right, a CCC occours in 10^20 individuals. Two CCCs occour in 10^40 individuals. The emergence of the cloroquine resistance is a CCC. The emergence of the SP resistance is another CCC. If your method of calculation is correct, there should be no Plasmodium falciparum cells on Earth which are resistant to both cloroquine, and SP, because that would need an insane amount of time.

So why is malaria still a deadly disease? If this is truly the edge of evolution, a single cell can be resistant to klorokin, or SP, but not to both, so treating the infection with both drugs should cure any infection.

Hovever these double resistant cells, harboring two CCC -s comapred to the wild type Plasmodium, have been detected, as early, as 1981, as malaria, that is resistant both to cloroquine and sulfadoxine-pyrimethamine, clearly harboring two CCC -s, comapred to the wild type, easily crossing the edge of evolution in Behe's own model system in a few decades, not in billions of years.

The usual chance-measuring used by Behe, and every other creationist is simply flawed. Two, three, ten, one hundred, etc. mutations do not have to occour all at once. One mutation happens, stabilzes in the population, then another one happens, etc. This way the individuals needed for two CCCs to happen, if you consider Behe's numbers to be accurate, are not 10^40 cells, but 2X10^20. Which can clearly be observed in Plasmodium falciparum, as it took two decades to get to cloroquine resistance, and another two decades to et cloroquine and SP resistant cells. Your method of measuring the population needed for this result is false, it can be observed in nature.

lee_merrill said...

Lee: Actually, it's a mathematical point

Anonymous: No, it's not. They don't evolve simultaneously *to* stick together.

But that is not my point, you changed my point, and then showed that this new point was not mathematical.

> Gyalogtank: The emergence of the cloroquine resistance is a CCC. The emergence of the SP resistance is another CCC. If your method of calculation is correct, there should be no Plasmodium falciparum cells on Earth which are resistant to both cloroquine, and SP, because that would need an insane amount of time.

Well, not quite! If there is a selectable stop along the way, then indeed a four-mutation sequence can evolve in a reasonable amount of time. In this case, there is a selectable stop for both resistances, if enough people are given just one of these drugs.

> The usual chance-measuring used by Behe, and every other creationist is simply flawed. Two, three, ten, one hundred, etc. mutations do not have to occur all at once. One mutation happens, stabilizes in the population, then another one happens, etc.

Yes, with selectable stops along the way, you can get as far as you want. But Behe's edge is at four mutations before selection gets to kick in.

Regards,
Lee

karmaiko said...

The odds of a favourable mutation - one that you're looking for, which the author is - are much higher than the actual mutation rate. Thus while a specific mutation is very rare to find, mutations themselves are occurring naturally all the time.

lee_merrill said...

> karmaiko: The odds of a favourable mutation - one that you're looking for, which the author is - are much higher than the actual mutation rate.

The odds of a selectable mutation being fixed are certainly much higher. But there aren't different rates for favorable mutations that I know of.

> In this case, there is a selectable stop for both resistances, if enough people are given just one of these drugs.

Which by the way, gives a way to test Behe's edge, if only we can find two such drugs for one disease, and administer both together.

ID does make predictions! If this edge is correct, then were we to have developed and administered chloroquine / SP together, malaria would be done for.

Anonymous said...

"there aren't different rates for favorable mutations that I know of"

I wonder if we've inadvertently stumbled across something new to say in the theist mindset debate.

Theists are stuck on 'purpose'. I think that's the fundamental problem / mistake.

The problem is that once you've decided that there's an overriding purpose to reality ... well, logically, everything meets that purpose.

At one level, there's a mindset that in its craziest form says God sent Katrina to New Orleans because lesbians live there. But the 'mainstream' Christian would thank God for the good things in their life and that's *exactly* the same thought process in play, just put to less malevolent use. God's got the steering wheel.

You're a good Christian who makes money - it's a reward. Lose money, it's teaching humility. If someone does something Unchristian and gains reward - well, the true rewards are in heaven.

Christians learn that doublethink early, and learn to navigate it.

What's more insidious: if you think there's a universal purpose underwriting everything that happens ... well, everything has to be part of that process.

And Lee betrays that here.

'Favorable', in this context, for an atheist/materialist can only possibly work in retrospect. we judge it favorable *after* the event.

In this case, the two proteins develop independently, it's (essentially) a coincidence that when they bump into each other, they stick. The reason it looks so extraordinary is precisely because almost none of the billions of other combinations there are would stick.

There is no such thing as a favorable *mutation*, only a favorable *outcome* of that mutation. And that's an absolutely crucial distinction that anyone who believes in a divine plan just won't get.

Not because they haven't read enough theology or science, or whatever, it's that once you concede that, you've conceded any notion of the Christian God as worshiped by any modern Christian. God doesn't have a plan for your life, he hasn't got the steering wheel, he can, at best, just put a tick next to the right answers.

(Although, I note that in Genesis, God creates something and *only then* sees that it's good).

Lee's hit on something, though. Obviously it's the opposite of what he thought he'd hit:

If God's around, guiding the process then 'favorable mutations' should be more common than unfavorable ones. I mean, at its strictest, why would there ever be *one* 'unfavorable mutation'?

But if mutation is being guided, we ought to see the odds *before the fact* tipped towards a 'favorable mutation'.

lee_merrill said...

> Anonymous: If God's around, guiding the process then 'favorable mutations' should be more common than unfavorable ones. I mean, at its strictest, why would there ever be *one* 'unfavorable mutation'?

This seems to be the problem of evil question--you know, it's been considered! Lots and lots of consideration.

Anonymous said...

Lee Merrill's grasp of reality is tenuous at best. But he does love him some Behe thought!

lee_merrill said...

> Lee: ID does make predictions! If this edge is correct, then were we to have developed and administered chloroquine / SP together, malaria would be done for.

This it seems also implies that if we were to develop medicines in groups of say, six or more, and always administer them together--such that at least four mutations would be required to resist them all, then the viruses / bacteria couldn't get around that.

Anonymous said...

"This seems to be the problem of evil question"

Well, yes, to an extent. I know that it bothers you theists. Never seen the problem, myself.

But this is a measurable, testable way of seeing if God is 'improving' things. Not even in a moral sense, but simply in a functional sense. A designer, however inept, at least *attempts* improvement.

And this isn't what we observe, is it? The process isn't 'random', but it's certainly not 'purposeful', either.

*If* a new combination works, it thrives (by definition). But it's always after the fact. We see absolutely no trend towards combinations that *will* work.

If Behr's right - and I'll defer to others on that - about the statistical chance of something like that happening ... isn't that a staggeringly clear example that the system *isn't* trending towards this purpose?

And here's the point:

The sort of forensic 'how did we get to here?' approach is the scientific approach. The 'where are we going?' one is the religious. You literally have to have things exactly backwards for this sense of 'purpose' to work.

lee_merrill said...

> Anonymous: We see absolutely no trend towards combinations that *will* work.

Unless Behe's edge is correct, and interventions are needed to get to some greater level of complexity.

Ceballos said...

Great post Larry... I learn a lot of evolutionary genetics with your posts, even though I'm phd student in populations genetics... Thanks a lot.
Regarding Behe's opinion, I agree with all your comments. Since molecular machinery is the core of life, it's origin is very old and still very unknown. The problem is that even though we decode the genetic message, we dont know how really works. I whant to add that I was thinking that Behe is not considering Motoo Kimura's theory of neutralisim, in it's very core. In my opinion Behe may be a extremist selectionist, (Dobzhansky's follower). If the main mutations are neutral nature can test lots of possibilities... do you agree?. What about if molecular machinery are quantitative traits?
In my opinion these guys are just coping the old theory of William Paley. However, the knowledge fild of evolutionary genetics still evolving whithing the science framework. IDots dont know how science works... evolutionary genetics is in it's childhood. We need to answer lots of questions...

Ingo Brigandt said...

Section I of http://www.ualberta.ca/~brigandt/Critical_notice_Sober.pdf explains what is wrong with an argument like Behe's. See in particular pp. 7-8.

Gyalogtank said...

>lee_merrill: "Well, not quite! If there is a selectable stop along the way, then indeed a four-mutation sequence can evolve in a reasonable amount of time. "

I presume, that you have proof, that there were no "selectable stops" during the evolution of your model system. (What is your model system?)

lee_merrill said...

> Gyalogtank: What is your model system?

The main one in Edge of Evolution being new protein-protein interactions.

> Gyalogtank: I presume, that you have proof, that there were no "selectable stops" during the evolution of your model system.

That would be the discussion at hand here! I am maintaining that you and others are proposing different scenarios, changing the edge, and then demonstrating that evolution can get beyond.

If indeed new (as opposed to enhancing current weak) interactions require the modification of six or so amino acids in one location, before selection kicks in, then there are no apparent selectable stops along the way--with regard to the function of the new interaction.

And if proteins typically will interact in groups of six or more, then such a new interaction from scratch is unlikely indeed.

Gyalogtank said...

>lee_merrill:

If indeed new (as opposed to enhancing current weak) interactions require the modification of six or so amino acids in one location, before selection kicks in, then there are no apparent selectable stops along the way--with regard to the function of the new interaction.

And what is your model system? Two hypothetical proteins? If those two tproteins are actin and myosin, they do not need any point mutations to form a complex. And as I asked before, what do you mean by "interaction"? From what Kd dou you consider two proteins as they interact?


"And if proteins typically will interact in groups of six or more, then such a new interaction from scratch is unlikely indeed."

Why? Why can't a complex begin with one protein, then acquire another interactor partner, then another one, and so on?

lee_merrill said...

> Gyalogtank: And what is your model system? Two hypothetical proteins?

Yes, Behe speaks in general terms.

> And as I asked before, what do you mean by "interaction"? From what Kd do you consider two proteins as they interact?

So, for a general interaction, pick a typical value? Maybe 1 fM or so?

> Lee: And if proteins typically will interact in groups of six or more, then such a new interaction from scratch is unlikely indeed.
>
> Gyalogtank: Why? Why can't a complex begin with one protein, then acquire another interactor partner, then another one, and so on?

Certainly it's possible, but it seems we would not generally expect new clusters of interactions to have such a selectable path, which results in a new six-protein interaction.

For something to be biologically plausible, it needs to not only be possible, but with in a reasonable timeframe, probable.

Russ Abbott said...

I'm surprised not to see the word bricolage in any of the answers. It seems to me that in explaining evolution we have failed to explain the importance of cobbling together existing things to create something new.

The key to so much of evolution is the richness of the component collection out of which new things may be created. Most evolution is not a result of a linear sequence of mutations. It is a result of the creation of large collections of elements, which can combine in various ways. It's the size of the population of components that counteracts the improbability of any one combination forming on its own.

One of my favorite examples is the Lancet fluke. The probability that the world would evolve specifically to support the life cycle of the Lancet fluke is, of course, minuscule. On the other hand, an ID explanation would have to say that cattle, ants, and grass were designed so that the Lancet fluke could live the life it does. That's even more ridiculous.

The story of the Lancet fluke must include the pre-existence of cattle, ants, and grass. Even then its evolution seems quiet amazing. But it would not have happened at all without them. And certainly cattle, ants, and grass were clearly not "designed" for its benefit.

When we explain evolution to the general public I think we do ourselves a disservice unless we stress the importance of its bricolage-like nature.

Larry Moran said...

@Blue

I'm a big fan of bricolage or tinkering [Evolution as Tinkering].

I'd love to teach public school children the concept of Evolution by Accident. But first I have to teach 95% of my fellow scientists, beginning with evolutionary biologists who, by and large, seem remarkably resistant to the idea.

Gyalogtank said...

>lee_merrill:

"So, for a general interaction, pick a typical value? Maybe 1 fM or so?"

LOL! So, as you have surely read int he wikipedia article you have linked, the biotin-avidin protein complex is the strongest protein complex known to man, with a Kd of 1fM. It is not a typical value, it is the highest value ever measured. Congratulations, you have just invalidated all your previous arguments. You have clearly spoken of "protein interactions", when you have set the treshold of the "interaction" to a Kd value, that only two or three proteins have ever achieved.

So let's examine your statements:

"the great majority of proteins in the cell work in complexes of six or more. Far beyond that edge"

"If either of these statements is correct, then it would seem the edge is practical. "

Groups of six? Can you show me one single example of six proteins interacting with Kd=1fM or less? I don't know any.

So if your statement of the treshold for "protein-protein interactions" is valid, it directly cotradicts Behe's staement, as the vast majority of proteins do not form protein-protein interactions at all, their complexes are weaker than 1fM by several orders of magnitude. If Behe's argument is correct, you have clearly misjudged the Kd of the typical "protein-protein interaction". So where is the error?

lee_merrill said...

> It is not a typical value, it is the highest value ever measured.

Well, I meant to pick a typical value, so say 1 uM. Say 1 nM. Say "a typical value", and you can tell me what that might be.

> Can you show me one single example of six proteins interacting with Kd=1fM or less? I don't know any.

And that wasn't what I meant--I freely admit I'm not a biochemist.

Schenck said...

Thanks to Prof. Moran, acting very nicely in his role as instructor here. The question he proposed in his post may as well be a 'short response' question on one of his biochem finals, no?
To repeat my understanding of the answer, and repeat me if I am wrong, the answer is NOT that Behe is wrong because he calculated the probabilities wrong, and its NOT that the two mutations don't need to occur together to be advantageous, and its NOT the the single mutations are not deleterious. Those would be answers on the test marked 'incorrect'. Its that you CAN get away with a deleterious mutation. The misconception is that evolution is a super-machine that relentlessly destroys all vice and promotes all virtue. And the lesson that Prof. Moran wants to test if we have learned, is that Dr. Behe IS HARDLY the only person with that misconception.

And if you consider that there was objection within the scientific community to 'neutralism', this is even worse, because here 'deleterious' mutations are allowable, and infact they can be promoted throughout the population.

Biology is sloppy, its a sloppy tangled, muddy bank.

(see that, we're getting a U Toronto edumacation for free, how'd'ya like them apples)

Schenck said...

RE: 'if there's a designer, we'd expect a higher rate of advantageous mutations'.

It really does sound like that must be some sort of test for ID no? Sure, there could be neutral and deleterious mutations even if there was a designer, so long as he simply 'allowed it' (and of course, is even that really that simple???), but we'd still expect that, when there was a mutation, it'd tend to be advantageous.


Also, the type of protein connections that Behe is talking about in the above, these are SERIOUSLY common events, are they not? If they're beyond the edge of evolution, and require intervention by a designer, then it means that, god isn't just jumping into evolution at a few critical points, but is CONSTANTLY meddling with creation. He's putting as much effort into bacterial flagellum as Human Brains. Evolution says that we just happen to be created by similar processes (similar existence), ID says that we were on par with one another in the mind of the designer/creator. That's a HELLUVA lot worse; yes, there is a god, and he thinks no more of you than a bacterial flagellum. Or put another way, "evolutions's" assesment of you is 'neutral', god's assement, you're shit.


Also, returning to the idea that by looking at proteins and the like we can find evidence of supernatural design (and that we can really only find it there), that's pretty weird. I mean, it says that the designer CREATED the universe, and was able to make it SUPERFICIALLY look like it was natural. But for some reason didn't bother to make protein evolution look natural. I mean, are we saying that the designer sucks and couldn't make proteins systems look natural, or that he's deceitful? The Designer in ID constantly sounds like a deceitful supernatural being, it sounds like a Devil God more than anything else.

Gyalogtank said...

That is the reason why this is a major flaw in Behe's argument. He is a biochemist, and mysteriously he never clarifies, what he wrote an entire book about. His whole argument is based on probability, yet he never states where he got his numbers from, only that he measured the probability of "protein protein interactions" occouring. And he has to know, he is an expert in this particular field. That's why I said earlier, that these are random numbers, Behe does not tell how he came to them.


What's the problem with this? Here in Hungary on the lottery you play five numbers from one to ninety. You can calculate the exact odds of winning, if you play once. You can calculate the odds if one, two, three, six, or ten numbers has to match with random numbers, and these odds differ considerably. For instance if you have to guess one number, one in every 90 lottery tickets will win.

These odds are reflected by the number of winners. Last week ~75000 people managed to guess two numbers, 2364 players guessed three numbers right, and 30 players managed to guess four numbers out of five. So, what are the chances of winning the lottery if you have to guess "some numbers"? How can you measure this without knowing how many numbers you have to guess?

Joe G said...

Larry:
Behe's version of the history of life requires a God who intervenes quite frequently to create specific mutations that are almost impossible to account for by random mutation.

No- a targeted search takes care of the need for intervention although intervention cannot be ruled out.

Also evidence, not rhetoric, is going to "win out".

Yes I understand that if a mutation is not fatal, just harmful, it can stick around for X generations.

However there was this recent peer-reviewed paper about getting two specific muatations and it didn't look good for you guys-

Waiting for Two Mutations: With Applications to Regulatory Sequence Evolution and the Limits of Darwinian Evolution