Evolution Disproved

[Administrator: This series of exchanges was culled from 7 pages of exchanges on the original file. Approximately three-fourths were insults and mockery to David Plaisted. The following series of exchanges was left when this and some extraneous material was removed. This is a good example of why this format is now being used for this board. ]

DAVID PLAISTED
Check out www.cs.unc.edu/~plaisted/ce/blocked.html
for a disproof of the theory of evolution.


SCOTT PAGE
Plaisted's 'disproof' fails utterly by the same token so many other 'disproofs' fail: 1. He engages in after-the-fact probabiliuty games 2. He lacks sufficient background knowledge to catch his errors in the first place.

Nonetheless, from computer scientist Plaisted's "evolution disproof" [sic]:

"In order to get these specific 34 mutations would require roughly 10^63 trials; this could be accomplished by a population of 10^30 individuals lasting for 10^33 generations, roughly speaking, or a population of 10^33 individuals lasting for 10^30 generations, et cetera. Clearly this requires an astronomically large population or a time much longer than available,or both."

Who caught the flaw in Plaisted's reasoning?
It is in the first sentence that I quoted:

"In order to get these specific 34 mutations..."

To show how irrelevant such calculations are, try this:

Calculate the odds of you finding a specific penny on a specific corner in a specific city at a specific time on a specific day in a specific year during specific weather conditions.
The odds are staggering, yet I just found a specific penny on a specific corner etc.... this very day.
Yet statistically, working backwards, I can easily disprove it.



DAVID PLAISTED
Actually this specific issue is handled quite directly in the article and my assumptions are even somewhat too favorable to the theory of evolution.


SCOTT PAGE
It is directly handled by using the same faulty logic. Once a mutation occurs, you are handling it as though it was specified.
For example, you talk about "the right hydrogen bonding", "every one of the specified[emphasis mine] 34 sites..."

You are taking an end product and working backwards, under the erroneous assumption that that specific protein is THE one, that it is necessary as-is, etc. You write:
"Thus one can expect most mutations that change an amino acid to be deleterious...", after referring to Crow's statement regarding mutations that can be phenotypically detected (was he referring to genes inlfuencing morphology or something else?). Tell me then - which cytochrome c found in the animal kingdom is the 'good' one? that is, since the sequence of cytochrome c's vary greatley across lineages, and they all seem to work, which ones should have been removed from a population by now?
In additon, aside form the occasional reference within the text (such as the second hand quote from ReMine), you supply no bibliography and a large number of your assumptions and 'conclusions' are citationless.

Lets say that you are correct. How then can you explain 'within kind' variation in only a few thousand years? For this variation would be due to the same factors that you imply cannot account for evolution... right?


DAVID PLAISTED
I'm not sure I understand your point -- my article gives a clear justification for saying that a specific set of 34 mutations is needed. But there is a valid related point which may be what you were getting at. My article assumes that the mutation process is evaluated on the basis of being able to reach a single new protein fold. Actually there could be many folds. This is a detailed part of the argument, but anyway, as far as we know there are only a small number of folds, possibly less than 1000. Thus we are only talking about a factor of 1000. In addition, some folds will be easier to get than others; the easier ones will dominate because the harder ones will be so much less probable. On the basis of a random distribution of folds in the "mutation distance space" one would expect only a small minority of them to be the easiest ones to get. And of these, an even smaller set would be beneficial to the organism. So it does not change the argument much to assume that there is just one fold that needs to be reached.

Oh yes, in kind variation can be explained by mutations that do not affect the shape of the protein, and also by different combinations of genes, recombination, epigenetics, gene duplications, and probably other factors.


THE BARBARIAN
It appears that Plaisted has rediscovered Hoyle's Folly.

Perhaps he'd be interested in thinking about the number of genes in humans, and then the genomes of his great-great grandparents.

Using his "proof", we can show that Plaisted is so unlikely that he is statistically impossible.

So now that he has proven that he and evolution are impossible, how are we to explain those speciations, fossils, biochemical and genetic results, and one slightly off-center web site?


DAVID PLAISTED
I see your point, but to me this does not appear to be related to my article. Maybe if you could explain it more clearly? The possible protein folds appear to be determined by the laws of physics and the question is how they could be generated by evolution. Another thing people may not realize is that all the cytochrome C etc in different organisms has the same shape. What my article is concerned with is not the detailed amino acid sequences, which have an element of randomness, apparently, and to which your argument might apply, but their shape.

THE BARBARIAN
Suppose a series of new enzymes evolves to catalyze a potential food substrate in a species of bacteria. The new enzymes obey the laws of physics in their folding. Why would we expect it to be any different?


Another thing people may not realize is that all the cytochrome C etc in different organisms has the same shape. What my article is concerned with is not the detailed amino acid sequences, which have an element of randomness, apparently, and to which your argument might apply, but their shape.

David, cytochrome c has the same shape in all organisms, because it has the same activity in all organisms. It is essential in glycolosis. The shape of the molecule is dictated by its activity. While some minor folding can be different, cytochrome c would lose its effectiveness if the active sites were different.


DAVID PLAISTED
David, cytochrome c has the same shape in all organisms, because it has the same activity in all organisms….

The problem is that the protein folds are separated by many mutations, to our best knowledge. In between is a wasteland of reduced fitness. Evolution always climbs the fitness landscape, or at least does not descend it. Therefore it cannot cross from one peak to another. The mechanism proposed to get around this has to do with "useless genes" and this is the subject of my paper. Typical beneficial mutations (if there are such) just change an amino acid and this requires only one mutation, so it does not require evolution to cross a fitness valley.

Harmful mutations tend to be eliminated from the population and beneficial ones tend to be retained, as well as some of the neutral ones. It is possible that harmful mutations could accumulate and render a population or individual less fit temporarily. However this will eventually result in a higher death rate for the population. Anyway, this has all been treated in population genetics studies and formal models have been developed to describe what happens. What we have to deal with is the probability of evolution behaving in a certain way and the probability of crossing a wide fitness valley is very low.


THE BARBARIAN
The problem is that the protein folds are separated by many mutations, to our best knowledge.

But where in the biochemical phylogeny of cytochrome c, are there many mutations? How many mutations do you think there can be in a molecule that size? Why do you suppose that there are some constant regions that rarely or never are changed?


In between is a wasteland of reduced fitness.

Show me. Between which two forms is there this "wasteland", and how do you know that?


Evolution always climbs the fitness landscape, or at least does not descend it. Therefore it cannot cross from one peak to another.

First, since all cytochrome c seems to have about the same ability to catalyze reactions, there seems to be no "fitness peak" with regard to the existing variations. Second, you are very wrong in thinking that a population cannot descend from a fitness peak. Indeed, you seem to be assuming that "fitness peaks" are a constant, and do not move about themselves. This they also do. A nice description of the process could be found in most introductory texts on evolution.

Changes in cytochrome c happen to be "just one amino acid". And there is no evidence whatever that such change involves a trek across a fitness valley, even if such "valleys" were a constant.


DAVID PLAISTED
There's a miscommunication here. Differences between cytochrome C in different organisms do not involve a change of shape. Thus there is no fitness valley to cross. The problem is how evolution could generate proteins having new shapes. This would involve many mutations to change from one shape to another and in between the protein would be useless. You should read the paper for details.


THE BARBARIAN
Dave, would you be interested in some research that shows proteins being changed to entirely new functions?

One way that the "fitness valley" can be avoided, is by duplicate genes. One of several to many copies can be modified, while the rest serve their original function. In this way, there is no "fitness valley".

But we know for a fact that useful new proteins can evolve, because we have seen it happen. So, we end up wondering whether we should go with the "new proteins are impossible" theory or the reality that they do evolve.


RUFUS ATTICUS
Evolution always climbs the fitness landscape, or at least does not descend it. Therefore it cannot cross from one peak to another.

You are 100% wrong. You seem to be using Wright's work without reading the entire volume. Populations can and do shift between "adaptive peaks" by genetic drift. Evolution is not all about positive selection. Selection does not always climb the hill either. I have personal experience with a frequency-dependent selection model in which the fitness of a population can decrease from natural selection alone. I hope you will fix your argument to take this into account.


MR BEN
This is where you make your fundamental mistake in your reasoning: The probability of evolution producing a particular pattern is not simply the naive number of permutations of that pattern, it is the cumulative probability of a mutation at each stage producing a better protein.

By your reverse probability reconing, I should not exist, as the probability of my parent meeting each other, falling in love, conceiving me, not to mention the specific random genes that I am made of, are astronomically huge.
But probability can only be worked in the forward direction. How probable is it that somebody 'like' myself would be born. Quite probable actually.

By the same token, applying probability forward on your 34 genes.. we can se that at any given stage (from 1 to 34), the probability that the odd beneficial mutation would come along to advance the utility of the sequence and eventually produce something 'like' the final result is quite high. That doesn't mean that if we did it again, we'd get the same 34 pairs. That would be very very unlikely (just like the chances of a genetic twin brother of myself being conceived by my parents), but we easily get another protein 20-40 that would do more or less the same thing.

So, you must ask yourself.. what are the chances that any mutation on a given genome in any protein will produce a better adapted individual. The answer is small, but large enough to happen all the time by chance.


DAVID PLAISTED
To Barbarian: Yes, I believe proteins can assume new functions by mutations. But this does not create proteins of new shapes.
But I would be interested in your examples anyway.

You say that the fitness valley problem can be avoided by duplicate genes. Yes, this is a very good observation and it is exactly this possibility that my article discusses in detail.

To RufusAtticus: All my argument assumes is that mutations that are harmful are eliminated from a population. It could happen that genetic drift would even make a whole population decrease in fitness since harmful mutations might spread to the whole population before they were eliminated. But this would require a small population and a very high probability of harmful mutation, it seems to me. Actually one can even have error catastrophe if the mutation rate is too high and the whole population could die out. In that case of course one would not have proteins of new shapes evolving either so it would not affect my argument. The question is how proteins of new shapes could evolve. But your frequency-dependent selection model sounds interesting -- is it in the literature anywhere or is there some way I can learn more about it?


To Mr. Ben: I believe I answered your objection in an earlier reply.


THE BARBARIAN
Yes, I believe proteins can assume new functions by mutations. But this does not create proteins of new shapes.

That would have to happen, if it had a new function. Maybe not a large change, but a change. What keeps that from accumulating small changes? We also know that happens.


But I would be interested in your examples anyway.

"Evolution on a Petri Dish : The evolved B-galactosidase system as a model for studying acquisitive evolution in the lab", Barry G Hall, Evolutionary Biology (1982) #15, pg 85-150.


RUFUSATTICUS
Rich Meagher has done some work on the evolution of the actin gene complex in plants. The genetic phylogenies indicate multiple gene duplication events with sequental diversifications. All gene families are basically formed the same way.

Molecular evidence from comparing the "same" gene from multiple species indicates that not every residue evolves at the same rate. Concerved regions, like active sites, don't evolve much. Initial evolution of the protein has basically optimized it. Protein evolution can also occur because of evironmental shifts. The old protein no longer functions "optimally" (but still functions), so some mutants can arise that are more fit than the old one.

You also need to remember that "less fit" is not equivilent with harmful.


But your frequency-dependent selection model sounds interesting -- is it in the literature anywhere or is there some way I can learn more about it?

The work that I am directly a part of is in preperation. (My advisor is in NZ now writing it up.) It is based on some earlier work by her. I can't recall off hand if she fleshed out the possible trajectories in that paper but here is the PubMed reference. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_ uids=2341034&dopt=Abstract

I honestly don't know of any additional FDS models that show the same effect, but I suspect they exist. You might want to try a literature search.


DAVID PLAISTED
Can a different selection model help protein evolution? I can't see how it could. Someone referred to "frequency based selection" and it took a while for that to sink in. The idea is that some mutations may be preferred based on their frequency, if for example individuals preferred mates that were unusual in some way. But I've thought about this and can't think of any reasonable selection model that would materially help the evolution of new protein shapes.

I think the largest misunderstanding has to do with protein folding. People really need to understand more about protein structure. I'll try to give some basics here. Proteins are composed of sequences of amino acids (the primary structure). Some parts of this sequence fold into regular structures called alpha helices and beta sheet (secondary structure), typically about half of a protein. Alpha helices are spirals and beta sheets are parallel or antiparallel and more or less straight. The rest of the protein is irregular. Then these structures fold together into some geometrical configuration called the tertiary structure, which is what I am concerned about. This has to do with how two alpha helices may rest against each other and so on. This tertiary structure can be called a "fold," though I'm not sure that is the exact use of this term. Now, the question is how new folds can arise. Less than a thousand folds are known and it is suspected that not many more can exist. (Another way of counting says about 10,000 but I think it means something else -- anyway the number is small). Each fold has restrictions on what kind of amino acid can appear at what position -- not just for the protein to be beneficial but simply for the fold to form. For example hydrophobic residues (amino acids) typically appear on the inside of a protein and other amino acids typically appear on the outside, but at least one amino acid can appear either place. Each fold has tight restrictions of this nature. If one replaces an amino acid by a similar one, the fold will probably not change, and this is the kind of changes in the different versions of cytochrome c. However, a random sequence of amino acids is highly unlikely to have the right kind of amino acids for _any_ fold and will become a useless glob that will probably react in harmful ways.

The question is where the folds came from since there are too many for all of them to have been present in the first reproducing system (it would be far too complex). Different folds are separated by many mutations -- to our knowledge, at least 70 percent of the amino acids have to change to change to another fold. As one travels an optimal sequence of mutations between folds, here is what happens: First the original fold becomes more and more unstable. Then the protein will not fold at all and will be a useless glob. Then the new fold will begin to form but will be highly unstable. Eventually the new fold will form and be stable. But even then there is no guarantee that it will benefit the organism. All the proteins on this path will be less useful than the original one, but the new fold may be just as useful. So how could evolution create a new fold? The duplicate gene idea is OK except the problem is that in the process of crossing from one fold to another, many other mutations will also occur that will randomize the rest of the protein and so at the end with very high probability one will have basically a random protein that will not form the desired fold anyway. And it will probably have many harmful added reactions. The details are in the paper and as far as I can see it is completely rigorous and refutes the theory of evolution. But if there is a mistake somewhere, no one in this discussion has pointed it out yet.

Thanks for your input, anyway. It has helped me to see where the argument can be misunderstood.


RUFUSATTICUS
If you really want to work the bugs out of your idea, why are you not consulting experts in protein folding and evolutionary biology?


MESK
I think the largest misunderstanding has to do with protein folding. People really need to understand more about protein structure. I'll try to give some basics here.

<snip explanation of protein folding>

This tertiary structure can be called a "fold," though I'm not sure that is the exact use of this term.


The two terms mean similar things, but are not precisely equivalent. The term "fold" has a broader meaning than "tertiary structure". Two proteins can have the same basic fold, but differ in the precise spatial arrangement of their structural elements (that is, their tertiary structure).


Now, the question is how new folds can arise. Less than a thousand folds are known and it is suspected that not many more can exist. (Another way of counting says about 10,000 but I think it means something else -- anyway the number is small).

The total number depends on how rigorously you define the term "fold", and the precise methods used to evaluate the data. Some estimates are based on the protein structures currently available from online databases, while others used theoretical calculations.

I did a quick Google search on the subject and found this site, which you might find useful:
Protein Origami: Exploring the world of protein folds http://www.cbs.umn.edu/class/fall2000/mimp/8006/review00/FOLDS.HTML


Each fold has restrictions on what kind of amino acid can appear at what position -- not just for the protein to be beneficial but simply for the fold to form. For example hydrophobic residues (amino acids) typically appear on the inside of a protein and other amino acids typically appear on the outside, but at least one amino acid can appear either place. Each fold has tight restrictions of this nature. If one replaces an amino acid by a similar one, the fold will probably not change, and this is the kind of changes in the different versions of cytochrome c.

That's true. Many amino acid differences between cytochrome c proteins are conservative (that is, the two amino acids are chemically similar), and these differences generally don't change the overall fold of the protein. For instance, the cytochrome c of humans and that of the protozoon Tetrahymena have the same basic fold, despite possessing less than 50% amino acid identity.

But don't be fooled into thinking that this means that the structure is somehow "protected" from amino acid changes. Many proteins possess certain residues which are absolutely crucial for correct folding, so that even a conservative change at these sites can change the overall fold of the protein. A similar effect can be achieved by multiple mutations at several different sites.


However, a random sequence of amino acids is highly unlikely to have the right kind of amino acids for _any_ fold and will become a useless glob that will probably react in harmful ways.

Sure, but evolution does not produce "random" proteins. It builds on pre-existing proteins that already possess defined tertiary structures. Changes in these proteins do not generally produce "useless globs", but can simply rearrange the preformed structural elements to generate different folds.


The question is where the folds came from since there are too many for all of them to have been present in the first reproducing system (it would be far too complex). Different folds are separated by many mutations -- to our knowledge, at least 70 percent of the amino acids have to change to change to another fold.

Well, not always. In general, proteins with greater than 30% sequence identity are likely to have the same basic fold. However, it is often possible to change a protein from one fold into another by altering only a few important residues.

As our understanding of the forces governing protein structure grows, researchers in the field of "protein engineering" are discovering precisely which amino acids need to be changed to produce specific changes in protein structure. This can include changes from one protein fold to an entirely different one.

[ January 10, 2002: Message edited by: Administrator ]

 

DAVID PLAISTED
Thanks for this good response. The question is how many mutations are needed to change a fold to a different one. If you just mean a minor change like an alpha helix resting at a slightly different place on another one, then just a few changes might suffice. But how does one explain new folds? I realize that just a single mutation can "break" a fold but what you get is then a protein which does not fold properly. The smallest case I know of is two amino acid changes to get a new fold, and this just changes a tiny part of the end of the protein into a slightly different fold, leaving the rest intact. This requires four point mutations. But anyway it is this very special case which is used as the basis of my calculation, even though it is being highly favorable to the theory of evolution since from what we know hundreds of mutations are needed to produce a new fold.



MILAN
Some mutations are harmful, some are neutral, some are beneficial. Natural selection discards the first, does not bother with the second, and favours the latter.

This is so obvious that the mechanisms that natural selection employs have been put to use in the laboratory in order to obtain enzymes with better catalytic activity than the naturally existing ones.

This is called directed evolution. This is a laboratory process whereby mechanisms employed during natural selection are employed at the molecular and single cell level to cause and then identify evolutionary adaptations to novel environmental challenges. This approach often includes modification of genetic sequences.

Since chemical reaction can occur very fast, multiple generations of molecules can be produced quickly.

Microorganisms such as bacteria can also reproduce rather fast allowing multiple generations of adaptation to be produced and studied in a short period of time.

Novel enzymes have been produced in this fashion, many of which have found industrial and biomedical applications.

Microorganisms have also been produced with the ability to thrive in previously inhospitable environments and may find use in waste clean up applications.

In the web you can find extensive information about these methodologies. Particularly interesting is the method of "gene shuffling".

See for example Crameri et al. Nature, 1998, 391(6664) pages 288-291 (or their webpages).


DAVID PLAISTED
This is a good point. However, the question is whether these novel enzymes have new protein folds -- if you are just joining this discussion, I'd suggest you read through the past responses to understand whatt hat means. It's also easier to get new enzymes that react with a chemical from the environment than proteins that react with other proteins in new and beneficial ways because protein-protein reactions require a very precise meshing of shape and chemical properties. Anyway, your post does not contradict my article, at least if I understand correctly.


MILAN
However, the question is whether these novel enzymes have new protein folds

Not really.
The primary structure of a protein determines the secondary and tertiary structures. Consequently, the formation of new protein folds, or whatever new form of secondary, tertiary or quaternary structures a mutated protein may display is a side effect of the primary structure.

If there are mutations in the primary structure, there will probably be changes in the higher structures. The question is whether the new protein is better, worse, or the same as the old one, or if it displays new properties. Not whether new folds are formed. New folds are a consequence of changes in the amino acid sequence.


SCOTT PAGE
Oh yes, in kind variation can be explained by mutations that do not affect the shape of the protein, and also by different combinations of genes, recombination, epigenetics, gene duplications, and probably other factors.

So, basically, evolution?


JOHN PAUL
According to the PBS series on 'evolution' copying errors & NS are what drives evolution. Copying errors are NOT recombinations or gene duplications. These genetic movements are controlled and precise. To use them in the ToE would be admitting Dr. Lee Spetner has a valid point. Do you want to do that Scott?


RUFUSATTICUS
If recombination is controlled and precise, why is unequal crossover observed?


MESK
Copying errors are NOT recombinations or gene duplications. These genetic movements are controlled and precise.

Say what?!? Exactly how is recombination "controlled and precise", John Paul? It's true that there are recombinational "hot spots" at specific chromosomal locations, but they are placed at random with respect to genes and unequal recombination between different chromosomal locations is not infrequent.
Are cancer, muscular dystrophy and congenital craniofacial deformities (to name but a few) likely to be due to a "controlled and precise" process?

And how on Earth is gene duplication "controlled and precise"? It's just unequal recombination that happens, by chance, to transfer an entire gene or genes to another section of the genome. The same process that causes gene duplication also causes more congenital diseases than you can poke a stick at through random duplication or deletion of complete or partial genes.

Oh, and don't try to turn it around by crying "see, mutations are harmful, therefore evolution is impossible". No one is denying that many mutations are harmful, and in any case it's irrelevant. All I want to hear is your justification for calling recombination and gene duplication "controlled and precise".


From David Plaisted:
Thanks for this good response. The question is how many mutations are needed to change a fold to a different one. If you just mean a minor change like an alpha helix resting at a slightly different place on another one, then just a few changes might suffice. But how does one explain new folds? I realize that just a single mutation can "break" a fold but what you get is then a protein which does not fold properly.

Not necessarily, by any means. It's true that some mutations result in completely dysfunctionally folded proteins (such as those that form the plaques seen in the brains of people with Huntingdon's or early-onset Alzheimer's diseases), but this certainly does not hold true for all proteins.

In general, proteins fold to form the lowest-energy conformation available, and in general that involves
(a) burying hydrophobic residues and
(b) forming ordered secondary structure, such as alpha-helices or beta-sheets, that maximise the number of non-covalent bonds between atoms in the protein. Now, while random proteins tend to fold relatively poorly, most proteins in living organisms possess certain features, such as numbers of hydrophobic residues, that allow them to fold constructively. Small changes to these proteins generally will not destroy these properties, so a mutated protein from a living organism is much more likely to fold constructively than a randomly generated peptide.

The precise tertiary structure of the protein is determined by the particular "folding pathway" that the protein follows to reach its low energy (folded) state. A mutation can change the final structure of a protein in two ways: either it can destabilise the normal fold of the protein, thus "pushing" it into a different low energy structure; or it can affect an early stage in the folding process, causing the protein to follow a different folding pathway and thus adopt a different final structure. Either way, because a protein in vivo is already optimised for the formation of compact tertiary structure, mutant forms of it are more likely to be able to form compact low energy structures - but the overall fold may be very different.


The smallest case I know of is two amino acid changes to get a new fold, and this just changes a tiny part of the end of the protein into a slightly different fold, leaving the rest intact. This requires four point mutations.

I assume this is from the Science article you cited on your web-page. You say that the new fold is only "slightly different" to the old fold, but in fact there is a significant structural change from beta-sheet to alpha-helix. Like it or not, this is a change from one distinct protein fold to another. Clearly the transition between folds can be achieved with as little as two amino acid substitutions.

I would also dispute that the transition must have required four point mutations. This is what the authors used to achieve the transition, but nature could almost certainly have been much more efficient.
To be more specific: the authors exchanged two amino acids as follows:
Original sequence: Gln-Phe-Asn-Leu-Arg-Trp
New sequence: Gln-Phe-Leu-Asn-Arg-Trp

But probably all that would be required to achieve the structural change would be to replace the asparagine (Asn) residue with any bulky hydrophobic amino acid, and the leucine (Leu) residue with any polar amino acid. Asn can be converted to the hydrophobic amino acid isoleucine (Ile) with only one point mutation, and Leu can be converted to glutamine (Gln) also with a single nucleotide substitution. It would of course require experimental confirmation, but I strongly suspect that these two point mutations would have almost exactly the same structural effects as the four used by the researchers.


But anyway it is this very special case which is used as the basis of my calculation, even though it is being highly favorable to the theory of evolution since from what we know hundreds of mutations are needed to produce a new fold.

I read through your calculations, and almost immediately spotted at least a dozen factual or logical errors. I don't think you're really to blame for this - I believe you said you don't have a background in biochemistry - but these will certainly need to be corrected if your "disproof" of evolution is to survive.
I'll list a few more glaring ones now, and perhaps do a more thorough review at a later date.

1. Your most obvious and pervasive error (which affects almost every single one of the arguments on your website) has already been pointed out by others in this thread: your use of the "probability in hindsight" fallacy. It is absolutely crucial that you understand why this type of argument is fallacious.

In hindsight, everything is statistically impossible. The probability of all of those exact events occurring in precisely the way necessary to make that leaf fall from that tree and land in that spot at that exact time are astronomically low - yet it occurred. It must be a miracle! Clearly it isn't a miracle, because leaves fall from trees all the time, but when you describe it in those terms it sure sounds like one. This is why probability in hindsight calculations are fallacious.

This type of calculation is exactly what you are doing throughout your essay. You say: In order for a protein to adopt a different fold its gene must undergo at least 34 mutations (this is an incorrect assumption, by the way, for reasons I'll go into later). Wow, look at the odds against those 34 exact mutations occurring at those precise genetic sites! It must be a miracle! But it's not a miracle - it's just a bad calculation.

To demonstrate this, assume (just for a moment) that evolution is true. Mutations happen. If they're harmful, they die out; if they're neutral, they drift; if they're useful, they are driven rapidly to fixation by natural selection. Imagine a gene that has undergone some selective changes over a period of time, just by random mutation and selection. If you look back at the probability of those exact mutations occurring, evolution would seem impossible - but it happened! It only seems impossible because you're looking at it in hindsight. And this is what your essay is doing - making the eminently possible sound impossible by looking at it in hindsight.

Your justification for insisting that such specific changes are necessary does not hold water. You cite a general guideline (hydrophobic residues are usually buried while hydrophilic residues are generally exposed to solvent) and try to use it to imply that only a pre-defined and specific set of mutations are required. This makes no sense, and certainly does not excuse you from the hindsight fallacy.

2. Even without the hindsight fallacy, your calculations are not analogous to evolutionary processes. If you think about it, your calculations give the probability of all 34 mutations occurring at once in a single "trial", from which you estimate how many "trials" would be needed for such an event to be likely. But this is nothing whatsoever like the way evolution works. Evolution occurs cumulatively, one selective mutation at a time, not all at once.

This false assumption makes an enormous difference to your calculations. You can test this (if you have a basic background in programming) by writing a program comparing the two processes. Define a target - say, a particular sequence of numbers. Then generate two random numbers of the same length. You can then "evolve" these two numbers to compare the process modelled by your calculation with the actual process of evolution.

So, for the first sequence you just keep generating random numbers until you stumble by chance upon the exact target sequence. This is the process modelled by your calculation. (Warning: for a target sequence of any respectable length, this process will take a bloody long time.)

For the second sequence, do this: take the random sequence, and generate a whole bunch of new sequences ("offspring") that differ from the "parent" at only a few sites. Compare each of the "children" to the target sequence. Figure out which one is most similar to that sequence, then use these as the "parent" for a new generation. Repeat the process until you end up with an exact match of the target.

Now add up the total number of sequences you had to generate in each case to obtain the target sequence. For the first process, which is the one your calculation models, the number will be massive.
For the second process, which is actually analogous to evolution, the number will be relatively tiny.
Hence, your calculation is a poor model of the evolutionary process.

(Note that this comparison assumes that each individual mutation is adaptive. I think this assumption is justified, and disagree with your assertion that intermediate mutations are neutral until the "final" mutation causes a structural change. Change from one fold to another is not a purely two-state process.
There is no reason why there cannot be intermediate structures which provide benefit to the organism.)

This point is really a little redundant since the hindsight fallacy renders your calculation (as well as this analogy) irrelevant to the probability of evolution - nonetheless, it is a good exercise in the importance of matching your calculation to the process you are trying to model.

3. You state the widely accepted guideline that proteins with less than 30% sequence identity are unlikely to have homologous structures, and then deduce from this that proteins with different folds must be more than 70% non-identical. This is erroneous for two reasons. Firstly, the "30% rule" is in fact a loose guideline - there are many, many exceptions, and one of the first things a structural biologist will tell you is not to rely on sequence homology alone when predicting protein structure. Secondly, you cannot equate the statement "proteins with less than 30% sequence identity are not structurally homologous" with the statement "proteins with more than 30% identity are structurally homologous".
This is bad logic as well as being factually incorrect.

4. You state that: "It seems that on the average one would have to change about half of the side chains from hydrophobic to non-hydrophobic and vice versa to get a new fold." Your 50% value is entirely arbitrary, you provide no evidence to support it, and given that the example you cite in your own essay requires only two amino acid exchanges to create a transition between folds it is also clearly incorrect.

That'll do for the moment. If I've accused you of a mistake you didn't actually commit, feel free to berate me.


JOHN PAUL
Mesk & Rufus,
I got that information from Dr. Spetner's book, Not By Chance. Here is a snippet from the ongoing debate Dr. Spetner is having with Dr. Max:
"Moreover, as I have noted in my book, the large mutations such as recombinations and transpositions are mediated by special enzymes and are executed with precision—not the sort of doings one would expect of events that were supposed to be the products of chance. Evolutionists chose the mechanism of randomness, by the way, because no one can think of any other way that beneficial mutations might occur in the absence of a law requiring them to occur. Genetic rearrangements may not be really random at all. They do not seem to qualify as the random mutations neo-Darwinists can invoke whenever needed for a population to escape from a local small adaptive maximum.

Evolutionists can argue, and rightly so, that we have no way of observing long series of mutations, since our observation time is limited to a relatively short interval. Our genetic observations over the past 100 years is thought to be more like a snapshot of evolution rather than a representative interval in which we can search for the required long series of changes. But our inability to observe such series cannot be used as a justification for the assumption that the series Darwinian theory requires indeed exist."


DAVID PLAISTED
Thanks Mesk, whoever you are. Even though posts like this are the most challenging, it is this kind of post I was looking for when posting to this group.

Let me just state some generalities, and then if there are any of your arguments left over not covered, I'll try to deal with them, too.

From Science Vol 277 11 July 1997 p. 179:
"pairs of natural proteins differing in up to 70% of their amino acid sequences virtually always fold up in to the same general 3D structure."

It follows that in order to get a different structure requires more than 70 percent difference in amino acid sequence, almost always. Now, there was a competition to reduce this. The best minds did all they could and finally they got a protein to fold into a different fold with _only_ 40 percent of the amino acids changed!! (Same reference) If you can do it better, then maybe you should tell them how, since the winner got $1000 for the accomplishment. If it is true that natural protein sequences can be changed with only a few amino acid changes then they should have known that and gotten a better figure.

Even the later Science article (9 April 1999) had to change 1/3 of the amino acids in the fold that was modified.
And this was highly favorable to a small number of mutations because:

1. It was on the outside of the protein, not buried.

2. It involved the ends of the protein, which are free and less constrained.

3. It was a small fold, only involving 12 amino acid residues.

4. It came from two identical polypeptide chains, meaning each mutation had a double effect, in effect halving the necessary mutations.

For folds inside a protein, changing the fold will also alter the surrounding geometry, requiring many more mutations. For folds not at the end of the polypeptide chain, modifying the fold will probably change the location of the other end of the fold and also require changes further on down the protein. Larger folds will also require more mutations.

It is possible that any bulky hydrophobic side chain would do as well as apargine, but not certain.

Even if you get a protein of a new shape the chance that it will benefit the organism is very small. For single mutations, the number of beneficial ones may be one in 1000 and is probably a lot smaller. For larger changes, the proportion of beneficial ones will be much much smaller. You and I both know that a new protein shape will not help an organism at all since many proteins need to work together. But for the sake of a rigorous argument suffice it to say that the chance wil be very small. The function of a protein is highly dependent on its shape so a new shape will require a new function.

Now, of all amino acid sequences, how many can conceivably fold properly? The largest estimate I have seen is that there are 80,000 protein families and it may be much smaller. There are at least 2 kinds of amino acids, hydrophobic and non-hydrophobic. Actually there are also amphiphilic (part hydrophobic and part hydrophilic) which often occupy interfacial positions (partly outside, partly inside the protein) (see Science vol. 278 3 Oct 1997 p. 80). Thus there are 3 kinds. But let's assume half of the amino acids can occupy a given site for each fold. Thus the chance that a random protein coded for by a gene of 1000 coding base pairs, that is, 333 amino acids, will fold into a given fold, is at most (1/2)^(333) or 2^(-333) which is about 10^(-100). If there are 10^5 protein classes then the chance that a random sequence of amino acids will fold properly into _any_ protein shape is at most about 10^(-95). Thus the more one randomizes a sequence of amino acids, the less likely it is to fold properly.

Now you claim that the proteins generated by evolution are somehow near each other so that it is likely to be able to go from one fold to another by mutations. But in the Science Vol 277 11 July 1997 p. 179 article, the best the investigators could do was come up with two folds with a 60 percent sequence similarity, not even restricting the proteins to be generated by evolution or be beneficial to the organism. If you have some special knowledge that one can do better, I would be interested where it comes from. An article quoted in my article estimates that there may be as few as 4000 protein classes determined by the laws of physics.

So my estimate of changing half of the amino acids to get a new fold was not bad -- one case required 1/3 change, with very favorable special conditions, and the other required a 40 percent change, while mostly one needs more than a 70 percent change.

The biggest issue is whether I am using the "probability in hindsight" fallacy. I am not calculating the probability of generating the particular proteins we see today, just the probability that evolution can generate _any_ new protein shape. This is different from the probability for mutations -- the probability that _some_ mutation will occur, over many generations, that will either be beneficial or at least not very harmful, is very high. I claim that the probability that _some_ new protein fold will evolve is very, very small. And as I said before, this probability decreases rapidly with the number of mutations needed, so it is basically the same as looking for the number of mutations t o the nearest new fold -- and of these, most will not benefit the organism. Thus the calculation is not changed much by assuming that there is a specific new fold that must be attained, if one also assumes that this fold requires a reasonable estimate smallest number of mutations to attain.

You say that my probability calculations require all 34 mutations to occur on a single trial. Actually, I made the situation more favorable to evolution by the way I did it. In reality it would take many generations to get enough mutations but I compressed this to benefit evolution. You say that each mutation could benefit the organism -- no, when you change the shape of a protein, the first fold becomes destabilized. Since natural proteins are stable, this means that unstable ones are harmful and are eliminated. Also, biologists themselves have proposed the gene duplication mechanism precisely because the intermediates are not beneficial to the organism.