Vowles and Amos - the issue of convergent DNA sequences

Welcome to this, the third thread to attempt to cover convergent tendencies in DNA and their implications for evolutionary theory and phylogenetics.In their PLOS paper Vowles and Amos show that small microsatellite repeat regions of DNA can influence local sequences and predispose them to certain mutations (Vowles and Amos, 2004). I brought this paper to Randman's attention in a previous thread in the context of the question of whether animals subject to similar selective pressures might not only exhibit convergent morphological changes but also convergent genetic changes.This topic initiated in the 'Is convergent evolution evidence against common descent?' and subsequently invaded the 'Macroevolution: Its all around us...' threads.There still seems to be quite a lot of ground to cover in terms of the exact implications of the Amos and Vowles paper and also in terms of the wider question of possible sources of convergence in DNA and the implications they might have on reconstructing phylogenies and our understanding of evolution in general.Could people please try and stay on topic and discuss the literature on genetic or morphological convergence and the relationship between the two, Haeckel already has a thread of his own and any other issues belong in other threads of their own.TTFN,WK

Thread moved here from the Proposed New Topics forum.

Let me start off with an unanswered question I asked in the other thread.You said:"Certainly it is highly unlikely that a protein coding region is going to be within the range of influence of a microsatellite..."I asked:So can we safely assume that phylogenies constructed from highly conserved and/or transcribed genomic regions will be largely unaffected by this potential source of error?Let me now further ask the following:Do you agree with Sasquatch that this convergent tendency discovered in the flanking regions is unlikely to result in any sequence that would have utility in transcription and, if so, can we possibly speculate on any evolutionary significance of the convergence?

Well, as Sasquatch pointed out, my statement was highly inaccurate. Given that microsatellites leading to effects on the local sequence can be as small as 'ACAC' there is no reson to suppose that they wouldn't occur in introns exons and all other genomic regions. Even a nine repeat sequence, the length producing the strongest measured effect, would only constitute 18 bp, equivalent to 6 amino acid residues.So we can't make that assumption for that reason although there are other possible reaons that we might make it, such as Sasquatch's point that selective pressure on a sequence might overcome any repeat related predisposition if the two favoured different bases. You might expect to see a trend in 3rd base positions of codons near to repeat elements, but I don't know how measurable such an effect would be. I don't know. Was that Sasquatch's position? I could certainly see it effecting upstream regulatory regions and I see no theoretical reason why it couldn't affect protein coding regions as well.TTFN,WK

An excellent point, and an interesting experiment. Let me try to clarify a bit, and see if my scanning of the paper gave me the right idea.What I was speculating is that the sort of convergent evolution described by V&A would not by itself produce coding sequence. I speculated such because it was my understanding that the bias produced in the microsatellite flanking regions resulted in convergence towards repetitive sequence; and such sequence by itself does not resemble gene sequence.I agree that the bias could influence gene or regulatory regions; my (intended) conjecture was that the bias alone would not consistently converge to functional sequence (which appeared to be the suggestion of randman in his discussion of a preprogrammed genome).I also maintain that phenotype-level selection pressure should trump the power of the mutational/stabilization bias (molecular-genetic-level selection); which is not to say that the two couldn't have an additive effect.

This much would seem fairly obvious, since the nature of the phenomena is limited to highly repetitive sequences, and therefore wouldn't have the required complexity for encoding proteins or performing higher-level regulatory functions. Oddly enough, it was as if he expected the process to create non-repetitive sequences or at least to have the potentiality for this -- and in accordance with some sort of pre-conceived plan -- where the information for this plan would exist elsewhere.This message has been edited by TimChase, 06-24-2005 04:03 PM

I wanted to clarify an important point for randman regarding the implications of the V&A paper, since he closed out the "Macroevo is all around us..." thread with complaints that all involved were denying non-random mutation, even though myself and others had already mentioned multiple times in the thread that DNA mutation was non-random.His claim was that Vowles & Amos had novelly demonstrated non-random mutation, and that their results thus required a modification of the Theory of Evolution.I'll repeat it though I already said it a few times in the other thread:It has been known for some time that DNA mutates in a non-random manner. If randman had critically read the V&A paper background info, he would have seen references that described non-random mutation over a decade ago. (Though I'm not sure when the first paper was published that demonstrated non-random mutation.)So the follow-up: what effect does non-random mutation as described have on the status of the Theory of Evolution?It seems another clarification is in order: That non-random mutation is not the same as programmed mutation, and simply refers to a bias in mutation. There remains a large random component to mutation of DNA.To put it another way: There is a 'T' at a particular position within a sequence. There is a molecular genetic bias in mutation, such that a mutation to 'A' is the most likely. However, mutations to 'C' & 'G' are still possible, and thus all potential base compositions are potentially available for selection.Sequence behavior as described by V&A results in a bias towards certain mutations, not inevitable convergence towards a certain sequence, nor complete prevention of certain mutations. If a base mutates from an A>C as a result of mutational bias, that 'C' is in no way immutable, and thus can mutate from a C>G despite the bias. If the C>G mutation happens to provide a strong fitness benefit to the organism, the 'G' will likely be maintained at that position by selection, despite the original mutation bias towards a 'C' base.I see no reason why the Theory of Evolution needs to be revised in light of biased (non-random) mutation of the genome, since a significant random component remains.

Additionally, one should properly view "mutations are random" as a kind of null hypothesis. When it is discovered that this null hypothesis fails in a certain domain, one should attempt to deliniate the boundaries of that domain -- which will oftentimes be suggested by the nature of the new empirical evidence. But one should not simply abandon the null hypothesis for all domains -- unless the nature of the new empirical evidence makes possible an alternate null hypothesis against which to test other hypotheses. This would simply be bad empirical science. And for an alternate null hypothesis to perform its role as a null hypothesis, it must be capable of making predictions which are falsifiable by means of experiments.This message has been edited by TimChase, 06-24-2005 05:20 PM

Also, I never learned that the phrase "mutations are random" meant that every mutation had an equal chance of occuring in the genome. This seems very unlikely. I was always taught that the "random" was with respect to the fitness of the organism. That is, a mutation isn't preprogrammed or directed to be beneficial. It just mutates without regard to fitness.A way to disprove this assumption would be to show multiple genomes all converging on some functional nucleotide sequence. As was pointed out above, "nonrandom mutation" doesn't mean "preprogrammed mutation." The latter, I would think, has been known for decades. The former has yet to demonstrated.[EDIT] changed "likely" to "unlikely" in first paragraphThis message has been edited by JustinC, 06-24-2005 09:08 PM

Alright.As an alternate way of stating this, I would say that the null hypothesis should be that mutations of the same size are equiprobable and that the rate of mutation (for mutations of a given size) is constant. (Obviously, this hypothesis could be refined as well -- but it is meant for the sake of an example.) But clearly this hypothesis has been replaced (or is in need of being replaced -- assuming such an alternate null hypothesis is available) in certain domains -- for example, when some organisms (such as viruses) are subject to environmental stress, the rate of mutation may climb by a factor of as much as 1000.In any case, I do not mean to suggest that this is currently what we should focus on -- identifying different null hypetheses for different domains -- only that this is the kind of approach that one might employ if one were to apply Statistics 101 directly to one's scientific methodology.In this case (Vowles and Amos), what we are interested in is a particular form of mutation which does not conform with the null hypothesis, but which may require qualitative analysis (e.g., causal, geometric, ...) prior to attempts to describe it by means of quantitative analysis -- where the description by means of a quantitative analysis would eventually involve the definition of alternate null hypothesis for a well-defined domain.Or am I missing something here?(Incidentally, I will bow-out and just "listen" as soon as needed -- as this isn't my specialty. But, if I feel I can contribute and it is not a problem, I will.)This message has been edited by TimChase, 06-24-2005 07:08 PM

Really? Because I find our discussions unfruitful because you don't understand your own models, and thus are ill-equipped to defend them. But that's not how rivers flow. Rivers flow in the least resistant path avaliable to them from point to point, until sometimes, heavy rains cause them to overflow their banks and new, less-resistant paths may be opened. Thus are things like oxbow lakes formed.How can you possibly expect to address the evidence for common descent when you clearly know so little about the natural world? But that's the thing. There is a pattern of similarities, and you've adopted convergence to try to explain it. How does the idea of DNA heading out in different directions explain why we see a pattern of hierarcheal similarity between the genomes of different species?You're trying to explain similarities among genomes by pointing out that the genomes of different species are diverging. How does that make any sense whatsoever? I did read them. The problem is that you didn't understand them.

Do you care to actually back that up? Specifically, can you address the cause of the "phenomena" as you call it, and address how this cause or these causes interact in other places as well as with micro-satellites?Your statement suggests you are well familiar with "the nature of the phenomena." So it should not be too hard for you to do. Really? No, I am just looking at the data without preconceived ideas on what to expect. I suggest you do the same. Could what you call non-repetitive sequences emerge, or coding DNA be affected, or have a potential for that? Your claim suggests this is not possible.The truth is on DNA I am not sure who is correct. The creationists say DNA mutations generally do not lead to enough variety, that the trend towards less repetitive sequences, if you want to put it that way, and maybe they are correct.But it's pretty clear to me that the level of understanding is expanding well, but still limited, as this study indicates. Imo, your dogmatic attitude is not reflective of scientific knowledge because if that was the case, you would be more cautious and admit that we don't really understand all of this yet, concerning convergent tendencies, but you take the opposite line.Wonder why?As far as the information for this plan, maybe you did not read the early hypothesis, that the local environment plays a factor. If this is a result of the chemical properties involved, and certain combinations will trend a certain way based on those properties, then there is indeed a design "elsewhere" in the sense of prior to the DNA forming that governs part of the process.Maybe you are just not reading the whole study or something?

Don't be an idiot. Anyone that has read my posts, especially on the convergent evolution thread, knows that I am aware that mutation is not purely random, but your claim suggests that somehow my speaking in those terms is improper and based on not reading the study.I suggest you reread the study because I am merely speaking in the same terms the paper does, and I have quoted that before. Frankly, I had decided to simply ignore this thread because it seemed like you and others deliberately were misrepresenting what I said, on purpose.Is that what is going on?Or, maybe you would care to also claim that the authors of the paper had not read it carefully enough too, since they refer to the random versus non-random argument?As far as other standard examples of non-randomness addressed in the paper, I read those, but you and others here are still ignoring the fact the writers feel that this is a significant departure from the assumption of "random" and independent mutations (their term not mine).Their emphasis is that there is some sort of directional factor involved and they even refer to the possibility of confounding phylogenetic analysis, something you guys have acted as if only an ignaramous could envision. It appears to me some here have just not read the paper, nor considered it's intent and findings, and yet have the gall to blast me for merely speaking of the paper in it's own terms.

This is your problem. Rather than look at each piece of evidence critically, you insist it must fit already into an existing overall hypothesis or theory, and thus your perception is colored. Hence your statement. I am not presenting "models." I am not approaching this, like you, with a preconceived idea of what the evidence supports. I don't have a believe first, understand later dogma when it comes to this.Personally, I think faith is a great thing, but I am not trying to take a faith approach to this area. By faith approach, I mean be taught what the conclusion is, and then view all the evidence as confirming that conclusion.Nope, I am just trying to look at what the evidence says. I see where examples of convergent DNA could cloud phylogenetic analysis, and apparently, I am not the only one. http://biology.plosjournals.org/perlserv/?request=get-doc...This message has been edited by randman, 06-27-2005 04:13 AM

That's actually a very interesting question. I remember writing an essay last year about this topic - I think there was a New Scientist article called "Genomes don't play dice" or something. Anyway, one thing that I found interesting was the idea that mutation isn't random, with respect to function. And it seems to be reasonably well supported. Not the idea that mutation is guided somehow, or that it "knows" where it's going. More like, in the genetic code how the 3rd base is often redundant so that mutation at that position won't affect protein function. Or how (in several species of bacteria at least) codon bias has worked so that many 2nd base mutations which don't result in the same amino acid at least result in one with similar chemical properties.Of course this is just all natural selection at work, just one step removed. Individuals less likely to produce deleterious mutations in their offspring would be at an advantage. Still, I thought it was an interesting take on things.