Predictable Evolution II
by MikeGeneAny interesting new study shows that evolution may be more predictable than many think:
Associate professor Yousif Shamoo and two students recently conducted experiments on a microbe, G. stearothermophilus, to see how it adapted to different environmental circumstances. In the experiment, the dominant strains of separate generations of the microbe ended up developing the same mutant gene in response to the same environmental hazards"¦"¦ The group then conducted the experiment again, and the same mutations developed. Thus, the experiment suggests that evolutionary development can be predicted, the researchers said.
"The duplicate study suggests that the pathways of molecular adaptation are reproducible and not highly variable under identical conditions," Shamoo said in a statement. "One of our most surprising findings is that an estimated 20 million point mutations gave rise to just six populations that were capable of vying for dominance. This suggests that very few molecular pathways are available for a specific molecular response."
All of this raises a very interesting question: To what degree can evolution be specified? Could evolution itself be an example of specified complexity?
Addendum: Another account of the same study.
May 22nd, 2006 at 11:02 am
[...] MikeGene asks: To what degree can evolution be specified? Could evolution itself be an example of specified complexity? [...]
Pingback by Heaven is not the sky » Blog Archive » Could evolution itself be intelligently designed? — May 22, 2006 @ 11:02 am
May 22nd, 2006 at 11:17 am
I don't see that the study (Why did you link to a news story?) is very interesting at all. We see the same degree of restrictions on (and reproducibility of) evolution in certain human tumors with identical amino acid substitutions in codon 12 of members of the Ras family.
Are you trying to insinuate that the mutations that were subjected to selection weren't occurring randomly? How do you equivocate between reproducibility/restriction and specification?
Semantically, the quote is gibberish–genes and mutations don't "develop."
Comment by Smokey — May 22, 2006 @ 11:17 am
May 22nd, 2006 at 11:50 am
Hi Smokey,
You may not find the study interesting, but anytime we can track evolution over 1500 generations, it is interesting to me.
So we're not dealing with an isolated phenomenon.
Nope. What we see is that the randomness of mutation is not terribly relevant, as the random search appears to be channeled.
There is no equivocation. I think the concept of front-loading evolution is rendered more and more plausible as more evidence surfaces that evolution is reproducible and under restriction. This then raises the question of specification. To what degree can evolution be specified? It's not a position. It's a research question.
Maybe. But then again, the teleological perspective allows us to begin thinking in terms of mutations that develop.
Comment by MikeGene — May 22, 2006 @ 11:50 am
May 22nd, 2006 at 12:06 pm
The bottom line here may be that because changes to amino acid sequence are immensely context-dependent (i.e. their effect depends extensively on which other mutations are already present), only a very small number of selectively important trajectories are likely to exist to high-fitness proteins.
Comment by Guts — May 22, 2006 @ 12:06 pm
May 22nd, 2006 at 12:14 pm
In response to the same environmental hazards?
Comment by Mung — May 22, 2006 @ 12:14 pm
May 22nd, 2006 at 1:04 pm
Exactly. Development and evolution as two different manifestations of the same process.
Comment by MatthewCromer — May 22, 2006 @ 1:04 pm
May 22nd, 2006 at 1:32 pm
Mike wrote:
"So we're not dealing with an isolated phenomenon."
That's why the study isn't very interesting based on the stuff you cited, which didn't include a manuscript.
"Nope. What we see is that the randomness of mutation is not terribly relevant, as the random search appears to be channeled."
I don't see any evidence for a search, and the channeling is merely selection.
"I think the concept of front-loading evolution is rendered more and more plausible as more evidence surfaces that evolution is reproducible and under restriction."
Obviously, but my question is why?
"To what degree can evolution be specified? It's not a position. It's a research question."
What experiments do you propose to test this apparent hypothesis? How would it apply to human tumors?
"Maybe. But then again, the teleological perspective allows us to begin thinking in terms of mutations that develop."
Not really. One can induce mutations by many methods, but mutations are the results, never the subjects of active verbs.
Mung asked re identical amino acid substitutions in codon 12 of members of the Ras family:
"In response to the same environmental hazards?"
No. Tumors in the same tissue with similar morphologies have the same substitutions. If you're interested in evolution, you should know about this. Bert Vogelstein will probably win the Nobel Prize for his work in this area, so it's anything but obscure.
Of course, discussing this in teleological terms says some very bad things about the motivations of the Designer.
Comment by Smokey — May 22, 2006 @ 1:32 pm
May 22nd, 2006 at 5:07 pm
I've seen a number of such studies. Many. One the themes of "Darwinism," as if that could actually be pinned down, is that all life evolves the same way: Universal Darwinism! (I've even seen it even in physics!)
The problem is if different life forms evolve via a different process, and not just evolve to be different via the same process, then "Darwinism" is false"”because it is "universal."
Stephen Jay Gould (?!) understood it a long time ago, a long time before biologists were publishing (many papers) on the differences between how the several "kingdoms" of life evolve.
I met Gould at a lecture he gave at UofC in ~1976, hawking his first book. We talked briefly, and he hawked some more. He contributed the Intro to Schindewolf's classic and told me I should read it. I did. I thought initially S was totally lost in his own theory of ammonite (morphological) evolution. He wanted to argue (and did) that that it followed a distinct pattern (peculiar to ammonites?), but that the same patterns applied to all life (orthogenetic, in short).
I thought, "You're nuts!" Then I thought, you're nuts enough to be right!
There is no reason to assume (no scientific) that all life forms evolve the same way! There is a reason why euks are different from proks, etc. They evolve by different rules of evolution.
There is no "common ancestor." A designer is not restricted to such nonsense He can invent any number of fundamentally different life forms to be and evolve to be anything he so chooses.
No designer is restricted to design according to our understanding of design (evolution).
Comment by Rock — May 22, 2006 @ 5:07 pm
May 22nd, 2006 at 5:21 pm
I should say that S was wrong. There is no reason to believe that all life forms evolve the same way as ammonites. And also, there is no reason to believe that proks and euks (etc.) evolve the same way, or anything that truly "evolves" evolves the same way.
This is what we have actually discovered! We've discovered"¦ not "evolution," but something else altogether"¦
What have we discovered?!
Comment by Rock — May 22, 2006 @ 5:21 pm
May 22nd, 2006 at 9:40 pm
Smokey:
What is deemed interesting is person-dependent. Smokey speaks for Smokey.
What data would you count as evidence for a search?
But that's insufficient. "Merely selection" means that a subset of mutations allowed these bacteria to survive under the high temps. Well, yeah. According to the report, "the researchers identified 343 unique strains, each of which contained one of just six variants of the critical gene." Why "just six" variants?
It's the noise problem that comes from trying to design through evolution.
If evolution behaves in a predictable fashion, perhaps one day we can actually predict what it will do. For example, with enough detail and a systems-level approach, it would have been neat if the researchers could have predicted that Q199R would be the most common variant and then found it. For now, I'll settle for less ambitious objectives. Yet first I have to lay out the hypothesis in more detail.
Comment by MikeGene — May 22, 2006 @ 9:40 pm
May 23rd, 2006 at 12:11 am
Smokey –
The channeling is not merely selection. The number of possible mutations in any one generation is staggering. For non-channeled mutations to occur so that certain ones can be predicted, they would have to occur at a rate which would cause error catastrophe in the cell.
The idea that they are channeled is very well-founded. You should see Barabara White's work on DNA secondary structures as a guiding mechanism for evolutionary change.
See:
http://baraminology.blogspot.c...
Comment by johnnyb — May 23, 2006 @ 12:11 am
May 23rd, 2006 at 12:44 am
That's just wrong. Especially with bacteria - in the course of experiments such as this, each and every position in the genome is easily sampled by random mutation, by mechanisms that plainly are not overwhelming or excessive of some error catastrophe limit. Heck, one can speed up the mutation rate that was "used" in this study by orders of magnitude without killing the population. Ask anyone who has worked wuth the phage Mu.
Of course, it helps to read the study, and realize that the "channeling" that ID proponents are so agog about is nothing more than a rather high selective advantage. Smokey is dead-on correct.
Comment by Art — May 23, 2006 @ 12:44 am
May 23rd, 2006 at 8:20 am
"That's just wrong. Especially with bacteria - in the course of experiments such as this, each and every position in the genome is easily sampled by random mutation"
But that's not what happens. A lot of these mutations are NOT single-base-pair mutations. Even if it was only one amino acid change, it sometimes requires multiple base pair changes to accomplish. Once you get into combinations, then there is no longer that kind of time available. For an example of a study on this, see Behe and Snokes. IIRC their study was for only a change in three amino acids.
Here is the paper
What is needed is a non-Darwinian mechanism where the genome is predisposed toward beneficial changes. Luck favors the prepared, darling.
Comment by johnnyb — May 23, 2006 @ 8:20 am
May 23rd, 2006 at 9:00 am
I'm talking about the paper by Shamoo and colleagues. The mutations of interest were single-base changes.
Such as the multiple changes documented, in real time, by Shamoo and colleagues.
Behe and Snokes studied a particular scenario in theory, and as others have pointed out, actually found that the resources in rather modest populations suffice to accomplish the theoretical "goal" laid out by Behe and Snokes.
Shamoo and his colleagues did not do a theoretical study. They actually performed an experiment with real organisms, and they found that the occurrences you are denying actually happened. To paraphrase an IDist whose credibility decreases with every passing day - we have the math and reality, I'll take reality every time.
The "predisposed" genome is one that has been littered with countless random changes, a tiny fraction of which may help in the stressful environment that triggered the mutational carpet-bombing.
The paper of interest in this thread actually sheds light into this as well (although not directly). I'd recommend actually reading it before commenting further.
Comment by Art — May 23, 2006 @ 9:00 am
May 23rd, 2006 at 12:03 pm
"Shamoo and his colleagues did not do a theoretical study. They actually performed an experiment with real organisms, and they found that the occurrences you are denying actually happened."
Bzzzzt. Wrong. What they found were changes. You are _assuming_ that it did a random sampling of single-base-pair mutations, but there is no evidence for that, and many theoretical reasons why it should not be so.
I do not have the paper, but nowhere in either the news description or the paper abstract does it say that the mutations were of a single base pair only.
This is a classic case of evolutionists simply _assuming_ that it is random mutations, instead of actually determining whether it is or not. In nearly every case, whenever someone is claiming that "random mutations" are the cause of a beneficial change, it is usually a label attached before the actual study of the mutations patterns were done. Read the Barbara Wright paper posted above, and you will see that she was able to account for nearly every observed mutation in the studies she looked at that were previously ascribed to "random mutation".
In this case, however, they actually did some study on it, and the abstract reports that only SIX alleles were observed to mutate. How on earth is only six mutating alleles random? THAT is directed variation. There's no other way to spin it.
Comment by johnnyb — May 23, 2006 @ 12:03 pm
May 23rd, 2006 at 2:43 pm
Mike wrote:
"What is deemed interesting is person-dependent. Smokey speaks for Smokey."
We were discussing the REASON why you deemed this interesting. There are mountains of data on recurrent mutations found in tumor evolution, which more than meets your criterion of 1500 generations.
"What data would you count as evidence for a search?"
The question is, what data suggested to you that there was anything resembling a search?
"But that's insufficient. "Merely selection" means that a subset of mutations allowed these bacteria to survive under the high temps. Well, yeah. According to the report, "the researchers identified 343 unique strains, each of which contained one of just six variants of the critical gene." Why "just six" variants?"
Because evolution is limited in its starting material, unlike design.
"It's the noise problem that comes from trying to design through evolution."
Please explain and apply this to the mountains of evidence for recurrent mutations we see in the real-time evolution of many tumors. Do they suggest design as well?
"If evolution behaves in a predictable fashion, perhaps one day we can actually predict what it will do."
But we can! If we design screens with abrupt changes in conditions to select for survival of new mutants, we predict a more limited number of mutants than we do for more gradual changes than were used in this study. In the latter case, we predict more complex changes, too.
"For example, with enough detail and a systems-level approach, it would have been neat if the researchers could have predicted that Q199R would be the most common variant and then found it. For now, I'll settle for less ambitious objectives. Yet first I have to lay out the hypothesis in more detail."
Color me skeptical.
Johnnyb wrote:
"For an example of a study on this, see Behe and Snokes. IIRC their study was for only a change in three amino acids."
Behe and Snokes didn't do an actual study.
Please answer a simple question about one of their underlying assumptions: I have a large protein family and have aligned their sequences. I know the crystal structure. If I change one of the conserved aa residues in the active site, contacting the substrate to a residue not found in nature, will I destroy or retain enzymatic activity, according to their assumptions?
"Bzzzzt. Wrong. What they found were changes."
I suspect that unlike you and Mike, Art actually read the paper.
"You are _assuming_ that it did a random sampling of single-base-pair mutations, but there is no evidence for that, and many theoretical reasons why it should not be so."
Straw man. Art pointed out that they FOUND single-base-pair substitutions.
"I do not have the paper, but nowhere in either the news description or the paper abstract does it say that the mutations were of a single base pair only."
So what? Art recommends reading the paper. I agree.
"In this case, however, they actually did some study on it, and the abstract reports that only SIX alleles were observed to mutate."
Your language makes no sense. The abstract reports that six mutant alleles were observed (after selection). Big difference.
"How on earth is only six mutating alleles random?"
Finding only six mutant alleles (not mutating alleles) after selection isn't random. You see, selection isn't random. Of the compound term RM+NS, only one of the component terms represents a random process.
" THAT is directed variation. There's no other way to spin it."
Perhaps you should read the paper before trying to spin it.
Comment by Smokey — May 23, 2006 @ 2:43 pm
May 23rd, 2006 at 3:11 pm
Johnny, when you read the abstract, did you notice this?
"Here, we have used a "weak link" method to favor variations in one gene…"
Comment by Smokey — May 23, 2006 @ 3:11 pm
May 23rd, 2006 at 4:52 pm
Comment by Mung — May 23, 2006 @ 4:52 pm
May 23rd, 2006 at 5:13 pm
"Please explain and apply this to the mountains of evidence for recurrent mutations we see in the real-time evolution of many tumors. Do they suggest design as well?"
Actually they do. There are two things involved.
1) A search doesn't hit 100% every time. It implies that there are false positives. A "search" will necessarily take you across space that isn't correct on the way to finding a correct answer. The question is, how well-organized is the search?
2) Most ID'ers think that the abilities of the genome are decreasing, not increasing, due to truly random mutation. Imperfections _accumulate_ and cause damage both to the system AND to the system's change mechanisms. Some mutations can put recombination signals in the wrong place and cause faulty changes to occur. In ID, there is a very big distinction between truly random mutations and "recombination mutation" for lack of a better term. The former slowly decaying the genome and its recombination process, and the latter assisting the genome in adapting.
"Art pointed out that they FOUND single-base-pair substitutions"
I'd like to see the relevant passages from the paper.
"Finding only six mutant alleles (not mutating alleles) after selection isn't random. You see, selection isn't random."
But it isn't magical either. According to RM+NS the genome should be littered with selectively neutral mutations after such an event in nearly every gene.
"Perhaps you should read the paper before trying to spin it."
I'd love to. Perhaps Art can send me a PDF: johnnyb@eskimo.com. However, abstracts are usually good enough to determine the content of the paper, and the abstract seems to support what I'm saying. I don't have the $30 to waste on purchasing it, and Interlibrary Loan takes 3 weeks. I may buy the full issue, simply for the 3D renderings of non-homologous end-joining (something I've taken a recent interest in), but again that would take several weeks to arrive.
Comment by johnnyb — May 23, 2006 @ 5:13 pm
May 23rd, 2006 at 7:38 pm
johnnyb wrote:
"Actually they do."
Tumors are designed? Why?
"1) A search doesn't hit 100% every time."
There's no evidence suggesting a search.
"It implies that there are false positives. A "search" will necessarily take you across space that isn't correct on the way to finding a correct answer."
A) Is there a single correct answer?
B) If the search covered a lot of space (or proteins were designed), shouldn't the existing proteins be distributed throughout sequence space? Is that prediction consistent with the data?
"The question is, how well-organized is the search?"
What search?
"2) Most ID'ers think that the abilities of the genome are decreasing, not increasing, due to truly random mutation."
What happens after selection? Are the heavily-mutated genomes of tumor cells passed on to offspring?
"Imperfections _accumulate_ and cause damage both to the system AND to the system's change mechanisms. Some mutations can put recombination signals in the wrong place and cause faulty changes to occur."
Then the organism carrying them is less fit.
"In ID, there is a very big distinction between truly random mutations and "recombination mutation" for lack of a better term. The former slowly decaying the genome and its recombination process, and the latter assisting the genome in adapting."
In ID, there's no experimentation nor evidence. Thinking isn't enough for science. There's that pesky doing part.
I wrote, "Finding only six mutant alleles (not mutating alleles) after selection isn't random. You see, selection isn't random."
"But it [selection, which isn't random] isn't magical either."
But I never claimed that it was. It's powerful, but not magical.
"According to RM+NS the genome should be littered with selectively neutral mutations after such an event in nearly every gene."
Dead wrong (pun intended). They were looking at the ones that survived selection. The sample size is tiny. Every gene would be hit in the population, not an individual.
Comment by Smokey — May 23, 2006 @ 7:38 pm
May 23rd, 2006 at 7:45 pm
Smokey:
Excellent. Could you please provide a couple of recent reviews that will help me better explore this mountain?
But that doesn't really answer my question - Why "just six" variants?"
Comment by MikeGene — May 23, 2006 @ 7:45 pm
May 23rd, 2006 at 8:06 pm
MikeGene asks:
What's wrong with the authors' explanation? After all, it gets to the heart of the matter, and explains why the only "channeling" is natural selection.
Comment by Art — May 23, 2006 @ 8:06 pm
May 23rd, 2006 at 8:10 pm
So if they had recovered 48 different sites that could be mutated to impart increased fitness, that wouldn't have been natural selection?
Comment by MikeGene — May 23, 2006 @ 8:10 pm
May 23rd, 2006 at 8:17 pm
LOL
"I haven't read the paper, but I know the authors are wrong." Must be a Post-Wedge World thing.
That version of RM+NS may be taught by creationists, but it is a very poor substitute for what biologists really learn.
Lemmee guess, johhnyb, you haven't really read any popgen either, have you.
Not likely.
Your proclivity to comment (inaccurately) on studies that you haven't read says it all. I'm content to leave our discussion there.
Comment by Art — May 23, 2006 @ 8:17 pm
May 23rd, 2006 at 11:35 pm
What does this have to do with anything?
From the paper in question:
Put briefly, the regime of temperature change tended to select for alleles that provided a very strong selective advantage. Not surprisingly, these are few. The authors indicate that a more gradual regime would have yielded more variants.
"Channeling" = strong selection. That's about it.
(The irony that is MikeGene's posting of an article that illustrates Dawkins' "Climbing Mount Improbable" theme should be pointed out. There, I did it.)
Comment by Art — May 23, 2006 @ 11:35 pm
May 24th, 2006 at 12:46 am
"Excellent. Could you please provide a couple of recent reviews that will help me better explore this mountain?"
Colorectal and prostate.
"But that doesn't really answer my question - Why "just six" variants?"
Art explained it very well.
Comment by Smokey — May 24, 2006 @ 12:46 am
May 24th, 2006 at 9:05 am
Art:
You said the only "channeling" is natural selection to explain the fact that only six variant sites generated a beneficial mutation. But the same explanation would be cited if 48 variant sites had generated a beneficial mutation. So your explanation did not explain why "just six."
This has been partially rectified with your quote from the article. You write, "Put briefly, the regime of temperature change tended to select for alleles that provided a very strong selective advantage. Not surprisingly, these are few. The authors indicate that a more gradual regime would have yielded more variants. "Channeling" = strong selection. That's about it."
Well, now we've changed something. Mere natural selection has now become "strong" selection. Or is it "very strong" selection? It's interesting that you assert, "not surprisingly, there are few" when the researchers themselves write, "Therefore, during adaptation, the accessible mutational pathways toward increased fitness are surprisingly small." It's a theme the lead researcher emphasizes in the press release: ""One of our most surprising findings is that an estimated 20 million point mutations gave rise to just six populations that were capable of vying for dominance," said lead researcher Yousif Shamoo, associate professor of biochemistry and cell biology."
Okay, so raising the temperature a half degree Celsius every 50-or-so generations is "very strong" selection. It would be nice to see data from the same experiment conducted over several months, even a year, with even more gradual increases. Do you think there is a linear relationship between the strength of the selection pressure and the number of sites that churn out beneficial mutations?
Given that strong selection is a key, this raises interesting possibilities. In fact, Smokey offers one himself: "If we design screens with abrupt changes in conditions to select for survival of new mutants, we predict a more limited number of mutants than we do for more gradual changes than were used in this study." Thus, can life itself initiate strong selection and abrupt changes in order to channel its own evolution? Y'see, it still doesn't cut it (for me) to say that channeling = strong selection and that's all there is to it. Why does strong selection channel?
As for Dawkins' metaphor, yes, the study shows there are surprisingly few steep paths up that mountain. Thus, to channel evolution, one thing to do is rig the system to look for steep paths.
Comment by MikeGene — May 24, 2006 @ 9:05 am
May 24th, 2006 at 10:13 am
Smokey,
Thanks for the articles.
Comment by MikeGene — May 24, 2006 @ 10:13 am
May 24th, 2006 at 10:56 am
I think it's an excellent question worthy of reseach. For example, the balance that must exist between mutation rate, error correction, and viability or avoidance of error catastrophe. Also the balance that must exist between mutation rate and evolution itself. Not enough mutation = no evolution.
More parameters for the anthropic principle?
Comment by Mung — May 24, 2006 @ 10:56 am
May 24th, 2006 at 11:06 am
MikeGene says:
I've still yet to see any such evidence that is not better explained by common descent alone than by front-loading.
Comment by Odd Digit — May 24, 2006 @ 11:06 am
May 24th, 2006 at 11:21 am
Also, it seems to me that "strong selection" would cause more mutations, not less. Modern neo-Darwinism does allow the cell to change rates of mutation. Therefore, it would have to change fast. Likewise, if the changes were not focused, it would cause more mutations in more alleles, not fewer.
Odd Digit —
Front-loading _is_ common descent. Front-loading is an idea of the origin-of-life, not anything past that.
Comment by johnnyb — May 24, 2006 @ 11:21 am
May 24th, 2006 at 12:55 pm
Mike asked:
"Well, now we've changed something. Mere natural selection has now become "strong" selection. Or is it "very strong" selection?"
It's artificial selection, remember?
"It's a theme the lead researcher emphasizes in the press release:…"
That's the problem with using press releases as a substitute for analyzing the data. Have you ever contrasted a press release with the carefully-qualified conclusions of a good manuscript?
"Do you think there is a linear relationship between the strength of the selection pressure and the number of sites that churn out beneficial mutations?"
I can't answer for Art, but I will note that I can't think of a single linear relationship in biology. I would predict that it would be roughly inverse exponential with a sharp cutoff at death.
johnnyb Says:
"Also, it seems to me that "strong selection" would cause more mutations, not less. Modern neo-Darwinism…"
As opposed to ancient neo-Darwinism or modern archaeo-Darwinism? Your obsession with using Darwin's name to label people has robbed you of the ability to write clearly.
"… does allow the cell to change rates of mutation."
In all cases? In the case of increased temperature?
"Therefore, it would have to change fast. Likewise, if the changes were not focused, it would cause more mutations in more alleles, not fewer."
Mutations occur in genes, not alleles. Please don't use terms like "allele" if you don't know what they mean. Were you trying to hypothesize that 'it would cause the recovery of more mutant alleles…" If so, that wouldn't follow, because the mutant alleles are only being recovered after selection. You seem to have a problem with this, which illustrates why only reading the abstract before drawing conclusions is unwise.
Comment by Smokey — May 24, 2006 @ 12:55 pm
May 24th, 2006 at 7:11 pm
So what's a mutant allele?
Comment by Mung — May 24, 2006 @ 7:11 pm
March 4th, 2008 at 5:21 pm
[...] Apparently evolution is more predictable than we previously thought (I don't have access to the paper yet, but you can read a discussion on it here and decide for yourself if I am representing the paper appropriately). [...]
Pingback by Roundup of Interesting Things I Don’t Have Time to Blog About and Roundup of Interesting Things I Don’t Have Time to Blog About — March 4, 2008 @ 5:21 pm