|
Author
|
Topic: Convergence or Divergence?
|
Jay
Member
Member # 268
|
posted 09. July 2002 11:46
Hi Frances,
Frances: "YOu are right, horizontal gene transfer is surely going to complicate the picture"
I propose adding another main theme to this thread. That is, how we infer horizontal gene transfer.
The domains and phyla are rife with examples of gene sequences that seem out of place phylogenetically. In fact, pieces of genes, whole genes, and whole gene systems are transferred all over the place, seemingly at will. This mix-and-match of genes between the domains is so prevalent that Woese proposed that perhaps life really is paraphyletic, and went on to point out that a gene phylogeny is just that - a phylogeny for *that* specific gene, and not necessarily a phylogeny for the organism.
Even among seemingly stable things like informational genes, pieces or whole parts are shared in such a way that makes the phyla and the domains appear truly chimeric in their makeup. It is completely against the common descent idea of a nested hierarchy.
Now, in nature we do observe some very limited instances of transfer of genes, with some viruses and a few bacteria doing some limited amounts of this. But, when we look at the 'HGT' in the domains and phyla, it is rather clear that we are looking at a different animal in terms of the degree of mixing and matching of genes. This degree of 'exchange' in the phyla and domains is such that it even gives Doolittle the feeling that relationships at this level are not so much variation around a common, core phylogenetic theme, but rather, indications that there is no core theme! ...Again, I think that we are asking the wrong question here in asking how they are related by descent! We need a new way of looking at these relationships.
But, we still see in the literature a sure feeling that the reason that we see this remarkable pattern of mixing and matching is because of gene transfer. But as we asked for homology, so we ask now, how is the inference of HGT made? And how are competing scenarios ruled out?
In asessing HGT, we do not know:
1.) The theoretical chances that the transfer of gene X could actually happen... what are the chances that it would have the proper methylation pattern to be accepted, that it would insert into a good spot, that it would be regulated properly, that it would actually give a selective advantage over the existing system, etc...
2.) The competing chance that the the 'transferred' gene is really convergent
3.) The competing chance that the reason we see all of this mixing and matching is because the organisms were built in a different way, i.e., not due to common descent and RM&NS.
So... as with our homology asessment, we often have almost no clue what our probabilities are for our favored scenario, or against competing scenarios. Yet the inference is still made.
So now we're up to two naturalistic inferences that are repeatedly made in spite of the fact that they have essentially no empirical evidence to favor them over other scenarios.
And I'll point out that investigating the nature of this gene sharing could be a great boon for ID theory. It may well serve as a convenient marker to suggest independent design for a certain class (i.e. when we find instances of certain types of gene sharing, we can infer possible common design for this class and the class that shares the 'transferred' gene).
ID theory would benefit greatly from finding marking points to show when and where design events happened. 'HGT' may turn out to be a tool in accomplishing this!
Thanks,
jay
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 09. July 2002 11:50
I find it interesting that Mike objects to my designer-centric approach when in fact he uses a similar approach when stating that the designer possessed a human-like intelligence but with access to a superior base of knowledge and advanced technology. But we still lack an understanding of the restrictions of such a designer or its methodology. Does it design using natural law, homology analogy? How do we recognize its design? ID is not only tentative it is without meaning unless we understand its mechanism. Mike claims that ID has a 3rd option but so does science, namely analogy, or convergence. Independent origins for the bacteria and the eukaryotes still remains as a possibility, whether or not it was designed... Well that will always remain a 'possibility' the question is can we find any supporting evidence...
If Mike uses the argument that contingency was rejected then surely he should also accept the rejection of analogy. But he himself showed the probabilities of a GTP motif to be no evidence against chance, certainly not for design which requires a far smaller bound (see Dembski).
Jay points out that HGT leaves many unanswered questions and suggests that ID theory (whatever that may be) may provide for more answers. I fail to see how ID theory would address issues that are not addressed by science. So now gene sharing may be 'evidence of common descent'? Perhaps it was the bacteria that did the design? A bit tongue in cheek but how does Jay propose to infer common design from this? What are these design events and the marking points? Why should we consider them to be design events? More importantly what does ID have to offer that science does not have presently?
So lets assume the following scenario, we find that there were conclusively two separate origins for life. How do we determine if there was ID or merely a natural origin?
What is a marker for common design? In fact how is common design limited? I still see no proposed mechanisms that would open ID to scientific investigation. Mike proposes a human like intelligence but more advanced what does this explain though, how does this limit the powers of ID? In fact how is this different from the Raelian hypothesis?
In the end science works by proposing a working hypothesis which can be tested. I have provided references to several testable and tested ideas such as a viral DNA origin. An overall phylogeny may help us determine the relationships between eukaryotes, archeaotes, and bacteria. Perhaps their differences are irreconcilable, perhaps there was some 'front loading' but was this designed intelligently or merely by nature and at what moment was it front loaded? Should it have been front loaded at OOL or at the level of the Big Bang? Why do we need 'front loading' as a hypothesis? Dembski needs it as a possibility to explain the origin of CSI but is this a necessity? I do not believe it is.
Perhaps we could explore what ID has to bring to the table that regular science does not? Does science ignore the possibility of a human like intelligence as the seeder of life? Does it prohibit it? Can it not address such?
[ 09 July 2002, 12:07: Message edited by: Frances ]
IP: Logged
|
|
Jay
Member
Member # 268
|
posted 09. July 2002 16:07
Hi Frances,
Cripes! I think that's the most questions I've ever seen in one reply!
Frances: "Perhaps we could explore what ID has to bring to the table that regular science does not?"
Simple. It has a new perspective that may help make better sense of some of these apparent contradictions in the data (e.g. it looks too different to have been homologous, but too similar to be convergent).
You 'regular science' may very well be throwing out a priori some good new ways of looking at the data.
And as for why I think that 'gene sharing' may be a marker for ID, well, the short answer is to ask what we mean by gene sharing. The current perspective takes this to mean actual physical transfer of information from one extant organism to another, with a heavy reliance on chance and selection to make this happen and stick.
What I propose is that we may be looking at this whole thing backwards. I propose that for many of these classifications of organisms, we're not looking at related organisms that share some genes by HGT, but rather, we are looking at separate organisms that were built separately via common, separate design, and also happen to share many common genes.
Common design means that we would likely expect the phylogenetic trees to run aground and give us no overall phylogenetic pattern, as this is the wrong perspective. It also means that between these organisms types, we might expect to find similar, analogous systems, since during common design, the designer(s) may have chosen to use a similar *idea* (like the ring clamps) in both systems, but to construct them separately. Here again, we would not expect these individual genes to ever show a real phylogenetic pattern.
So overall, breaks in organismal as well as individual gene phylogeny is a good *starting point* to suspect common design. It then prods me to look further into the specifics to see if the pattern continues to suggest it. And for the domains, and their ring clamps and cytoskeletons, the hypothesis seems to be holding up well.
It's a whole new way of looking at the same data, which will lead us to ask different, and hopefully better questions (hopefully leading to things like cleaning up alignments!) This is what the design perspective offers.
Thanks,
jay
IP: Logged
|
|
Jay
Member
Member # 268
|
posted 09. July 2002 16:39
** New Subtopic - The Use of IC As An Identifier of Common Design With Sequence Similarity
There... just felt like adding a title ![[Smile]](smile.gif)
We've been debating so far whether systems that share significant sequence similarity might be truly separate systems due to convergence/common design. One of the ways that we have been discussing is by looking for systems that fall outside of the likely reach of both homology and convergence, based on their sequence alignments, barriers to evolution, etc...
I propose that another simple way that we can parse between real homology and convergence/common design, and perhaps even slice between convergence and common design, is to use the inference of IC, and to examine the components of an IC system to see how they relate to each other. We do this to see if we might pick up on some common design themes within this one system.
For instance, if we examine the flagellum, we find that many of the components share sequence similarity. This is true for many of the big likely-to-be-IC machines. In fact, one of the arguments against IC -> ID was that many of the components look 'homolgous' to each other. However, rather than this being an IC -> ID liability, it may actually be a large benefit.
As we have pointed out already, this homology inference based on sequence similarity is quite weak, and may well completely cave in as the inferior explanation when we start dealing with components of IC systems. In other words, simply finding out that two genes within an IC system share sequence similarity is no reason to assume homology, and to then use this to argue against the system being IC (and ID).
If we take two similar genes from an IC system, we have the paradox that we:
1.) need both parts from the start
2.) that the parts look like they may be related, simply on the basis of sequence similarity.
From a naturalistic perspecive, this is hard to explain if indeed the system is IC. It doesn't make sense from this perspective to have two 'homologous' genes that were both needed at the same time. However, with our new design perspective, it is easy to view this as really no paradox at all, but rather, positive evidence that we have a designer with the foresight to assemble many different components into a functioning system, while at the same time, sharing sequence and motifs between some of the components.
So, for those who think that the IC inference is a dead end, I say hogwash! It may well be an excellent marker to spot common design with sequence similarity. And if we can find enough of these examples, it may give us an overall perspective on how and why sequence similarity is shared. It will also give us model systems to test any single gene common design detectors on.
In the end, IC may help lead to a single gene common design inference, which may lead to a more comprehensive theory of design, as well as providing a means to establish what really is homology, which may lead to practical benefits such as better drug design and a better understanding of protein folding due to better homology modeling... and who knows what else!
The ID perspective is a gold-mine of new, good ideas just waiting to be explored!
Thanks,
jay
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 09. July 2002 23:12
I asked Jay what ID has to bring to the table that regular science does not. Jay responds that it has a new prespective but that does not make it different from regular science. Of course in cases where the present hypothesis lacks in explanatory power, science will look for different hypotheses and the idea that organisms may be showing more than one common ancestry has been proposed by several people, most recently by Woese. So the idea of multiple common ancestors is hardly new. So I fail to see how ID as proposed by Jay could help us beyond what science is commonly doing?
But we need a hypothesis and an explanation why the hypothesis organisms were commonly designed by an intelligence is better than the hypothesis that organisms were commonly designed by nature? If we are allowing the idea of front loading then how are we sure that what we see is evidence of direct or indirect design? Would there be differences between front loading at OOL versus at the time of the Big Bang?
A break in phylogeny is good evidence that something else happened, but why common design? Are we not jumping to conclusions or premises that are unnecessary so far? What about the premise that there are independent origins?
Jay continues in a later posting to propose IC as an identifier of common design. But unless IC is a reliable indicator of ID, why should we conclude based on IC that there is evidence of common design when IC systems could have arisen naturally?
Jay suggests similarities in genes within an 'IC' sytem as evidence of ID but why is this? Would gene duplication followed by specialization not be a viable alternative for which we do seem to have independent evidence?
You are presuming that the IC system was IC from the start but that seems to be unwarranted. In fact there are quite plausible pathways in which ICness may arise through the loss of a 'scaffolding'. Behe has argued that ICness is hard to explain naturally and yet plausible explanations have been given so IC perse is not sufficient to infer design. Which means that we are back to the start, how do we find independent evidence of design?
Until we can limit the explanatory power of ID, that is explain how the IDer would go about designing almost anything would fit the ID mold. That does not make for a very productive hypothesis and runs the risk of becoming an ID of the gaps.
I do not share the optimism with Jay that the exciting times and research are due to an ID perspective. In fact I would argue that Jay's forms of ID are quite well dealt with by science as it is right now. Perhaps ID offers some new ideas but whether or not they are good? It seems far to early to guess.
In order for ID to deal with the many problems of the design inference, it should focus on defining the limitations of design. Mike has chosen to limit his designers to be human like but more advanced. Does this help us explain the limitations of these designers? How do we detect the design if we cannot describe the limitations of the hypothesis?
In fact, in Dembski's design inference, we need to understand the probabilities of chance (which includes regularities lately it seems) but in order to understand these probabilities we need to understand the causal history of these structures. But is that not how science works already? Delving deeper into our understanding of the structures? Perhaps what we need is a probability approach to intelligent design as well but that would mean we need to understand its causal history of the 'creation of the structure'.
In my perhaps somewhat pessimistic view I do not believe that ID can develop a scientific theory based on eliminative arguments such as IC and ID.
But I love to be shown wrong.
IP: Logged
|
|
Mike Gene
Member
Member # 149
|
posted 10. July 2002 01:13
I'm up against my own self-imposed 50 Rule , so I will make this my last contribution to this thread.
Bill Dembski writes the following is his latest reply to Richard Wein:
quote:
Francis Bacon advised "Read not to contradict but to weigh and consider."
It appears this advice is also not being followed in this thread. Frances crafted his latest reply to me less than ten minutes after reading it (he posted 12 minutes after my posting and I'm giving him 3-4 minutes to type up his reply). In my opinion, one really can't weigh and consider something in ten minutes. And this is unfortunate. In the same reply to Wein, Dembski also framed this whole issue very nicely:
quote:
.... I want to start by reviewing my project. It starts by asking a straightforward question: "If an intelligence were involved in the occurrence of some event or the formation of some object, and if we had no direct evidence of such an intelligence's activity, how could we know that an intelligence was involved at all?"
It would seem that this is one of the most fascinating questions we can ask. Just how might we infer design without independent evidence of the designer or designer-in-action? It not only touches on origins, but epistemology (in some ways, it's similar to the age-old philosophical inquiry about detecting the existence of other minds). What's more, this would seem to be a most difficult and challenging question. These ancient origin events are fuzzy enough (as can be seen by Frances inability to provide any clear evidence for homology between ftsZ and tubulin). But to add teleological explanations to the fuzz must give many a severe headache.
However, it would seem that engaging such a fascinating, difficult, and challenging topic would draw serious attempts from more people who would want to address such a line of thought divorced from the all the baggage that comes up with "The Debate." Yet it is my subjective impression that most critics of intelligent design would rather stonewall Dembski's question than engage it. This is what the designer-centric approach (DC approach) is all about. Since Frances' lead-off in his reply to me suggests he is a little confused about this, let me repost how I spelled this out:
quote:
[it] assumes that independent knowledge of the designer is necessary to carry out a design inference.
The DC approach answers Dembski's question by replying, "It can't be done!" That is, we can't ever hope to detect/infer design without having independent knowledge of the designers and their methods. Of course, I think the designer-centric approach is confused. Specifically, while independent knowledge of the designers and their methods will make a design inference easier, and also make it easier to reach a consensus, I do not think such knowledge is necessary. As I have been trying to stress, it all depends on what one is trying to extract from the data of the physical world. Are we trying to establish certainty and reach consensus? If so, perhaps there is something solid to the designer-centric approach. Are we trying to conduct a tentative investigation? If so, the designer-centric approach is not necessary. I explained why in my last posting and no counter-argument has been forwarded.
The main point is how we would go about inferring design without independent evidence of the designers and their methods. Those in the designer-centric school of thought have given up without ever really trying. Yet there are consequence to the designer-centric approach that I'm afraid its proponents have not seriously pondered.
One consequence is that it adds confusion to the debate. To see this, consider a couple of Frances' assertions that appear to have been spawned from the DC approach:
quote:
ID is not only tentative it is without meaning unless we understand its mechanism..... I still see no proposed mechanisms that would open ID to scientific investigation.
The focus is on "mechanism," which is understandable from a non-teleological mindset. But I suspect many read "mechanism" through non-teleological filters. To appreciate this, we really need to translate these claims to better capture what is being asserted:
quote:
ID is not only tentative it is without meaning unless we possess and understand the lab notes, protocols, recipes, blueprints, assembly technology, etc...... I still see no lab notes, protocols, recipes, blueprints, etc. that would open ID to scientific investigation.
So? Why anyone would think we need the actual lab notes, protocols, recipes, blueprints, assembly technology, etc. before we can engage in a productive investigation is beyond me. Heck, if we had that stuff, the "investigation" would probably be quite trivial and not the intellectually stimulating challenge that it is. The fun comes from thinking about ID without the luxury of having the lab notes, protocols, recipes, blueprints, assembly technology, along with first-hand experience of the designers-in-action.
But it gets deeper than this. Earlier in this thread, I pointed out (several times) that the truth of intelligent design does not entail we'd be able to access the lab notes, protocols, recipes, blueprints, assembly technology. The truth of ID is divorced from our ability to obtain such information. What's more, I can't think of a good reason why we should expect to obtain the lab notes, protocols, recipes, blueprints, assembly technology. Like I said:
quote:
But what if life was designed? Does this entail that we should be able to find "unambiguous evidence of a designer." Nope. In fact, I cannot think of one good reason why one would argue, "Since life was designed, it follows we should be able to get our hands on the blueprints or recipes." None.
Frances does not address this point. Yet it is an essential consideration that shapes the context (and thus explains why we talk past each other). The DC approach essentially boils down to this:
Mike's Point: If X is true, we don't expect to find/have A and B. (X= design of life, A = knowledge of the designers, B= knowledge of their methods)
The DC Approach: To show X is true, you need to produce/use A and B.
In other words, the DC approach demands that we focus our attention on things that don't follow from the truth of life's design. Why is this?
Is it a mere coincidence that most of those who champion the DC approach also just happen to be ID skeptics? Could it be the demand for A and B doesn't stem from a desire to investigate, but from a desire to be "convinced" while adopting a hyperskeptical stand? After all, if we had independent knowledge of the designers and their methods, the ID critic would be forced to accept design inferences (as they do with pyramids, Mt. Rushmore, etc.) The ID critics thus demand they be forced into an ID inference. The DC approach is simply a demand for epistemological evidence, rather than the ontological evidence that interests the investigator. Put simply, the DC approach is about convincing "skeptics" and not about investigating. The reason I bring all this up (again) is that the lion's share of Frances questions about designers and mechanisms are indebted to this DC approach.
But then what of Frances' question, "So lets assume the following scenario, we find that there were conclusively two separate origins for life. How do we determine if there was ID or merely a natural origin?"
I'm not going to list my suggestions in my last contribution to this thread. In time, and elsewhere. But in the meantime, we can say to Frances, "Yes, how do we determine if there was ID or merely a natural origin?" Apparently, Frances' answer is that we get our hands on the designers and their methods. Otherwise, we can't tell the difference and must appeal to game rules or metaphysics. But there is a profound implication to the thoroughly agnostic position Frances has embraced. It simply means that we have no reason to think that the origin accounts floated by non-teleologists intersect with truth. If Frances has no way of determining that these origin events were "natural" rather than artificial, then the origin accounts he entertains may very well be Myths (and I don't use this term negatively). And this is where the DC approach inevitably leads - even if Life was designed, we'd think of other stories to explain it. And this, in turn, casts a dark shadow on all his attempts to cite scientific authority. For as I have explained, all of his scientific arguments collapse for the simple reason that none of the conclusions were arrived at by trying to "determine if there was ID or merely a natural origin."
Think about this DC approach. It would have ID theorists expend all their energy looking for things not predicted to exist by the design of life (designers and their methods). It's basically a challenge to "convince the hyper-skeptic." And if design were true, it leads us away from this truth. I confess I can understand the superficial appeal of the DC approach, but deep down, it is thoroughly misguided and useless (even harmful) to a teleological investigation.
Getting back to Dembski's question, I thus reject the DC approach. I look for other ways. This doesn't mean that any ID inference is going to be clear. Or that it will successfully generate consensus. The DC approach would make things immensely easier in these regards if it could be run. But then my focus is not on convincing skeptics or generating consensus.
Finally, let me clean up some points raised by Frances.
quote:
If Mike uses the argument that contingency was rejected then surely he should also accept the rejection of analogy. But he himself showed the probabilities of a GTP motif to be no evidence against chance, certainly not for design which requires a far smaller bound (see Dembski).
I originally showed this in response to Frances' "evidence" of homology. That is, when I noted that it is structural similarity that was used to conclude homology, Frances raised the GTP motif as supporting evidence. Yet as I showed, this is not convincing supporting evidence. The big point here, after these 50-or-so exchanges, is that although Frances claims homology as "the best explanation" and cites scientific authority, we have seen that there is no compelling evidence/data to choose homology over the two other options before us.
Now, I did say that I do accept scientific opinion that allows us to reject coincidence. But I have always been upfront about relying on fuzzy thinking and intuition. As such, I can accept the fuzzy thinking and intuition that is behind the scientific community's rejection of coincidence as an explanation. The problem comes when some embrace the fuzzy thinking and intuition used to support non-teleological explanations and exclude teleological explanations because they employ fuzzy thinking and intuition.
Let me, in good faith, answer most of the questions:
What is a marker for common design?
In this case, something too similar to be accounted for by coincidence/convergence, yet too different to be accounted for by homology. And yes, right now one has to rely on fuzzy thinking here. Just as in science.
In fact how is common design limited?
A research question. I would hypothesize that the cellular context is what sets the limits. That is, the attempt to design different, yet similar cells, would bring design constraints into the picture. Firm answers to such questions may depend on advances in our own bioengineering, attempts to design novel cells/proteins, and even nanotechnology.
I still see no proposed mechanisms that would open ID to scientific investigation.
Irrelevant, as explained above.
Mike proposes a human like intelligence but more advanced what does this explain though, how does this limit the powers of ID?
The more advanced the intelligence, the more reason to reject examples of "bad design" as design. Think about it.
In fact how is this different from the Raelian hypothesis?
I'm not very familiar with this hypothesis. From my sketchy knowledge, this hypothesis is driven by religious/metaphysical reasoning and selectively chooses data to reinforce those beliefs. My proposal is a genuine attempt to explore/answer Dembski's question from a biological angle (I have no religious motivations for entertaining them). What's more, the Raelian's go too far in specifying the identity of their designer, as such specification does not follow from the data in the world. I'm disappointed (but not surprised) that I have to spell out these differences.
In the end science works by proposing a working hypothesis which can be tested.
I likewise have done this concerning my working hypothesis on the web page article I linked to. You may complain that my tests don't distinguish ID from "natural origins," but remember that you too failed to come up with tests/results that distinguished homology from convergence/coincidence and common design. You too are unable to distinguish "natural origins" from ID.
As for the questions about front-loading, I invite you to read previous answers and my postings I linked to. You are just repeating questions that have already been answered.
Perhaps we could explore what ID has to bring to the table that regular science does not? Does science ignore the possibility of a human like intelligence as the seeder of life? Does it prohibit it? Can it not address such?
Science cannot/does not consider teleological explanations for the first cells. Christian DeDuve, and many others, have spelled this out clearly in their writings. But perhaps the easiest way to appreciate this is that science does not have a method or test to either detect such a cause or rule out such a cause. Hundreds of OOL papers exist, but not one seriously addresses such a cause. ID thus brings another perspective and highlights different points of emphasis.
Anyway, as I said, the 50 Rule is in effect:
quote:
The 50 Rule - I will not participate in threads that expand beyond 50 postings. It has been my experience that such lengthy threads usually stray beyond the original topic and I often find it frustrating to check in on a thread that is discussing things not clearly related to the original topic. By the time a thread gets to 50 postings, it is usually time to fire up a new thread with a new (perhaps related) topic.
Clearly, what I wrote in March applies here, as we're now arguing about a mishmash of ID-related topics. But hey, it was fun while it lasted.
[ 10 July 2002, 01:20: Message edited by: Mike Gene ]
IP: Logged
|
|
Jay
Member
Member # 268
|
posted 10. July 2002 01:54
Hi All,
Frances,
Some of the questions you raise are quite good, however, I see that we are going in a loop here. I feel that Mike and I have answered as best as we can many of the questions that you have raised these dozens of times. I really like Mike's last reply, and feel that it sums up (better than I could) much of the problem here. Quite honestly, I'd say that over 80% of the stuff you've asked in your last reply was already asked several times, and answered several times. I get the strong impression that you aren't really reading what we're saying anymore.
One of the other underlying problems I see, is that you seem to intepret what Mike and I readily admit are tentative inferences to mean conclusive evidence, or stuff that would convince the hyper-skeptic, as Mike calls them. We have no intention of doing such things. If you wish to forever be the skeptic of these things, then fine. But we are hashing out new ideas, as well as pointing out to the interested reader just how fuzzy 'regular science' is when it comes to origins.
I wanted to really get across that 'regular science' is very fuzzy in regards to origins. Things like homology and HGT are very fuzzy inferences where very little is known about the mechanisms or probabilities - and I mean very little. It often comes down to just gawking at something and saying that it's too similar to not be homologous (a false dichotomoy, if I ever saw one! As though it's only a race between convergence and homology!). Their inferences are very poor on data, yet they still make and run with them. We intend to do the same thing with an ID inference.
You'll likely ask again why we should bother with this - well, for you the skeptic, it probably doesn't make any sense. If a naturalistic worldview suits you, then fine. But I've outlined many times why I like the ID inference, and don't intend to keep repeating myself.
Frances: "In order for ID to deal with the many problems of the design inference, it should focus on defining the limitations of design."
Yes, and that is precisely what we're trying to do. For instance, finding how to detect common design in organisms could well show us at what level (and even when) design happened. The topics on this thread are quite relevant to this question. I feel that you need to actually read and consider our answers again - I think we have already begun to answer some of your concerns.
Mike,
Don't leave!!! Well... do so if you must. But, I'd really like to hear your opinion on using IC to generate possible examples of proteins that share sequence similarity but are not related by descent. It's a new idea I've been kicking around, and may have lots of holes in it that I haven't considered. Alternatively, if you want to stick to your limits, e-mail me if you feel like it.
Thanks,
jay
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 10. July 2002 03:21
I am somewhat saddened by the tone used by Mike to respond to my postings. I was hoping that this group would be different from the usual 'personal' arguments.
Mike suggests that "In my opinion, one really can't weigh and consider something in ten minutes."
or "Since Frances' lead-off in his reply to me suggests he is a little confused about this, let me repost how I spelled this out:"
Let's start by quoting Bacon
"To know truly is to know by causes. "
Let me once again address Mike's claims. Let me once again point out that the Design Inference ala Dembski or IC ala Behe is eliminative, it provides ID with a special status of that what cannot be explained by chance. It tries to do this by showing that chance hypothesis are 'unlikely'. Since it does not propose an independent ID hypothesis it does not place chance versus ID but chance all by itself. That is, no attempt is made to show if ID might be more or less likely. Combine this with the claims that the filter is reliable, which according to Dembski means no false positives and one sees the problem.
To me the solution is obvious, the design hypothesis needs to provide us with a way to determine its probability. But how do we establish a probability of an event if we do not understand the causal history of the event? Without such knowledge there is no hope to detect design, certainly not without plenty of false positives.
What if science just claimed that natural forces explain our sun without making an attempt to provide for a hypothesis how these natural forces led to the sun shining brightly in our skies? What would be the value of such a hypothesis?
Does this mean that we cannot entertain the idea of ID such as proposed by Mike, a human like intelligence but more advanced? Sure, there are people who may consider this to be a like scenario but how is one going to propose a reasonable way to determine if this 'hypothesis' explains the observation if one does not understand the causal history of what is argued to have been designed? How is one going to propose design if one has no way of limiting what the 'designer' can do?
Mike discusses the sequence data and indeed by itself the data is not helpful in deciding homology versus analogy so we are stuck with independent origins or common origins. There are those who argue for independent origins, so it seems that science surely is addressing the possibilities raised by Intelligent Design, without necessarily jumping to conclusions about the 'designer' without some further evidence. In fact I have shown how people have proposed DNA viruses as a possible explanation of the origins of these genes. Such a hypothesis can be tested and rejected or we may find it to be quite convincing given the fact that the hypothesis provides us with mechanisms.
Perhaps Mike is right, we might not be able to detect design without further independent evidence, especially if perfectly good explanations exist.
And perhaps we are ignoring a perfectly good idea but science has not to deal in hypothetical 'what ifs' that have no bounds in their explanatory powers.
So far science has been quite succesful in closing the gaps of design, should we still allow design as a possibility? Sure, there is no a priori reason to reject it, and science does not do such but the question remains, is there a way science can resolve this issue.
My problem with Dembski's argument is that he suggests that an eliminative filter in which ID has a special status can detect design without false positives. It would perhaps be helpful if ID would either support its claims in that area or realize that the filter approach is not going to be helpful in infering design without false positives.
You claim that design centric people have given up trying, I find that an interesting argument that seems to be based on little more than a subjective interpretation of motives and actions. Science and scientists are always trying to find new hypotheses by what hypotheses could be generated from ID that would not follow from science as we know it right now?
SETI is attempting to detect design in the cosmos, have these scientists given up? Or are they realizing that they are looking for a signal which can be recognized as if it were sent by a human like intelligence. Since we understand how human like intelligence may communicate we may be able to detect such signals but how do we intend to detect signals of design in OOL, something we have not even achieved ourselves? Can we use ourselves as a model here to detect design in nature? It is even more complicated because there are several competing hypotheses for OOL.
Mike seems to realize the problems when stating that he does not expect independent evidence such as lab notes etc to be found. Mike claims that the truth of ID is divorced from this. But as I have argued, I believe that ID cannot make such claims, certainly not at this moment.
Mike might want to claim that one would not expect to find any evidence of methods or the designers but then I would like to know what ID has to contribute to science? What hypotheses can it raise that can be shown to have any relevance to reality? What are the limits of design without knowing something about the designers/methods?
Mike's suggestion that my interests (or in his words ID critics/DC proponents) may have no interest in investigating. But my statements should put such claims to rest. The issues of homology and analogy of these genes is fascinating but I see no reason to presume ID nor do I see any additional values in such a presumption.
Mike seems to be frustrated by the idea that ID scientists would spend their time looking for things not predicted by the design of life. My response to this is two fold. First that seems to be the inherent problem of the design inference combined with claims of no false positives, secondly what is/is not predicted by design of life is something that is not restricted in any sense.
When asked for a marker of design, Mike responds with a weak analogy of Dembski's filter, too similar to be coincidental, to different to be homologous. A non response since it also places design in a special position and presumes that coincidental/convergence and homology are the only alternatives. As I have shown, independent origins would explain the data, what about seeding from outer worlds? Another possible explanation, what about DNA viruses?
When asked about the limitations of design Mike responds, the more intelligent the less like bad design. That's the argument used against Intelligent Designers and examples of jury rigged 'design'. So how do we determine the level of intelligence ? By the quality of the design? But the quality of design determines the level of intelligence. So now we have a nice tautology at most.
Mike suggests that science cannot consider teleological explanations for the first cells. Why not I ask him? What makes first cells so different from detecting life or pyramids?
Perhaps Mike is right though, science often does not have ways to rule out or test teleology which makes a teleological approach hard to be scientific. But then what is teleology if it is not scientific? If Mike wants to explore philosophical origins of life as possible scenarios that's fine but if he believes that science should deal with issues with which it cannot deal without further data, then I wonder what ID has to contribute?
ID might be right but we may never know...
[ 10 July 2002, 14:27: Message edited by: Frances ]
IP: Logged
|
|
Jay
Member
Member # 268
|
posted 10. July 2002 15:35
Hi all,
Well, this thread has explored quite a few diverse topics, so I thought I'd take a pause and establish where I think that we are currently at. I believe that we have investigated and somewhat established that:
1.) The homology inference is very fuzzy, and often relies on no more than the simple observation that a certain protein is 'just too similar to not be related'. This inference does not take rigorous probability estimates into account as is often demanded of the ID inference.
2.) The inference of lateral gene transfer likewise is often no more than simple declaring that two genes are just too similar with respect to others to not have been somehow transferred natualistically. HGT makes the assumption that actual, physical transfer from one organism to the other is the reason that the two genes look so similar. But again, this inference does not rule out convergence or common design. Like the homology inference, it is based on the way things subjectively look to the researcher, rather than strong empirical data.
3.) In both of the above inferences, common design is never even considered, but rather, all choices are made within a naturalistic context even when that context is strained.
4.) I believe that the sliding ring clamps and the FtsZ/tubulin are good examples of high strain in the homology inference, and well illustrate the way in which the homology inference is still made even when it is a very strained and superficial interpretation (as it is here).
5.) The sliding clamps and FtsZ/tubulin fall into a sort of 'dead zone' where they don't really look homologous, but likewise do not look convergent. This apparent dead zone with high probability barriers on either side may well be a signal to us that we need other way of sharing these similarities. That third way could well be common design, as common design would fit these 'sharing common ideas yet truly separate' scenarios quite well.
6.) These two examples positively represent what could well be interpreted as *idea* sharing. This has strong design overtones. They seem to reflect a common engineering way of doing things, yet at the same time, defy an actual descent relationship.
7.) Another good place to look for instances of common design, this time when sequence is shared, may be in IC systems. Here, the homology inference is strained in a different way - by the fac that both parts were needed from the start.
8.) Learning to spot common design is useful. Homology modeling is becoming increasingly important, and it makes some of these same assumptions (possibly mistaken ones) that those who infer homology make - namely, that shared sequence means descent and divergence from a common ancestor. However, should some similar genes be the product of common design, our homology models will be wrong, since they make the wrong assumptions. So detecting common design vs. common descent would greatly aid in cleaning up these alignments and making our homology models better reflect reality. This can lead to all sorts of practical gains in knowledge.
9.) The ID infence may well help clean up the messy inference of homology. The ability to parse between common design and homology, and to still rule something homologous, will give us a more robust inference. And it will help deal with this 'dead zone' examples that seem to better fit with common design.
I believe that the ID inference is on its way to including single genes, and that doing this will help flesh out a more comprehensive Theory of Design, as well as benefitting the homology and HGT inferences by making them more rigorous.
Thanks,
jay
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 10. July 2002 22:43
I just read through this very long and interesting thread and many of the links. The poster Jay has done us the favor of summing up his points, so I, with some trepidation, will respond.
I think I mostly disagree with Jay and Mike Gene (not Mike Behe I take it as Frances presumably miswrote once in this thread), mostly because I think that the inference of homology/common descent (which are not necessarily the same thing, depending on definitions) made by all of those professional biologists and biochemists cited throughout the thread is far less fuzzy and squishy than Jay and Mike are suggesting.
Normally homology is equated with common descent by biologists. However, if common descent (e.g. between two proteins, or two phylogenetic domains) is exactly what is in dispute, as is the case in ID discussion such as this, I think it is probably better to return to the original definition given by (nonDarwinian, notably) Richard Owen.
Jonathan Wells lays out the history briefly here:
quote:
From at least the time of Aristotle, people who study living organisms have noted some remarkable similarities among very diverse creatures. Bats and butterflies are quite different from each other, yet both have wings to fly; bats fly and whales swim, yet the bones in a bat’s wing and a whale’s flipper are strikingly alike. The first kind of similarity involves different structures which perform the same function, and in 1843 anatomist Richard Owen called this “analogy.” In contrast, the second kind of similarity involves similar structures which perform different functions, and Owen called this “homology.” Owen (and other pre-Darwinian biologists) attributed homology to the existence of archetypes: biological structures are similar because they conform to pre-existing patterns. (Bowler, 1989; Panchen, 1994)
Now as Wells notes in the following passages, Darwin argued that Owen's homology was better explained by inheritance from a common ancestor than by archetypes. When the arguments for common descent (from homology among others, e.g. biogeography) were widely accepted, common ancestry of parts and homology became essentially two words for the same thing.
Unfortunately Wells rapidly forgets the distinction between Owen's homology and analogy, and begins making use of the much broader, vaguer term "similarity". I see this happening quite often in the thread also.
Here is my main point: there are different kinds of similarity. Owen found two broad classes of similarity in biology, and the key distinction between them is the relationship between the similarity of "parts" (organs, or by extension proteins) and the function those "parts."
Thus mere talking about "similarity" misses the point. When comparing two parts, you have to:
1) Identify the functions of those two parts (if any, there might not always be one)
2) Identify the similarities in "construction" between the two parts, and see if those similarities are necessitated by the similarity in function.
Thus there are a number of possible situations when comparing two parts. The two main ones are:
- Same function, but different "construction"
- Different function, but same construction
The first Owen called analogy, the second homology. His key insight was that there are similarities that cannot be explained by similarities in function.
Analogies, on the one can be explained by similarity in function. They could be caused by any function-sensitive process, e.g. a common designer, or e.g. by natural selection for the same function. A classic example is the eyes of vertebrates and cephalopods, which at a gross level are similar, but the similarity is simply that required by the physics required for having a working eye. In the details there are all kinds of profound differences. (I am leaving aside the recent discovery of a few shared developmental genes which indicates that there might be some very distant homologies between vertebrate & cephalopod eyes; those homologies do not specify the whole structure of the eye)
Owen's homologies, on the other hand, are those similarites which cannot be explained by similarity of function. They are similarities in excess of those required by similarity function. Thus homologies are easiest to identify when similarity of function is very low: any detailed similarity (I say detailed to exclude similarities explainable by chance) is unexplained by functional similarity, and thus needs a different explanation.
I would also argue that structures with a high degree of similarity of function, which have similarity in excess of that required to specify that similarity of function, are also homologous.
OK, I've beaten that to death I suspect. Here is my/Owen's point regarding homology: it is unexplained by common design for common function. Owen hypothesized "archetypes" to explain these similarities (and actually moved towards common ancestors himself according to this page), and Darwin proposed common ancestors.
The only real definition of "archetype" that I've seen is "common plan", generally in the Divine mind I suppose, although here a very rigorous definition becomes difficult. My point again is that the idea of "archetype"/common plan, according to Owen's logic, does not include similarity due to similarity of function. It is extra similarity.
OK, that's plenty for the moment.
I will next attempt to apply the definitions and conclusion of this general analysis of the term "homology" to the protein homology/domain common ancestry debate in this thread.
(It may take awhile though)
yersinia
PS: A reasonably coherant review of the standard biology-class account of this is here.
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 11. July 2002 00:58
So, in the previous post I defined homology and analogy independently from common descent. I think that the key reason biologists consider common ancestry a good explanation for homology is that the "archetype" explantion really doesn't explain much at all. Why would a designer employ a common plan that is (by definition, Owen's definition) including extra similarity beyond that needed for the similarity of function of the two parts in question? Particularly, why would the designer include extra similarity between parts with two different functions? There appears to be no good reason, particularly for a divine designer but presumably for any hyper-advanced designer capable of e.g. creating life. In any case, hypothesizing a reason for making use of this common plan would require making a hypothesis about the designer(s) and their capabilities/limitations/habits/whatever, and it is generally acknowledged that we have no way to test these hypotheses so they are at best unsupported.
(if the above argument is disputed, preferably let's do that in another thread; in any case, I want to point out that we can distinguish homology and analogy, independent of our explanation for the homology and analogy)
So, how do we establish (Owen-style) homology in the case of proteins? In my opinion there are several advantages in this case that make it even easier than in classical anatomical cases.
I regard the following premises as empirically established and widely accepted. If others dispute them, please be explicit why:
- For any given protein structure, there are a very large (huge) number of DNA/amino acid sequences that can produce that structure. The usual case is that amino acid sequences can show no more similarity than would be expected from randomly generated sequences, and still produce very similar tertiary/quaternary structures.
- For any given biochemical function, there are a large number of tertiary structures which can perform that function. Even if a general shape is required for e.g. a structural function, there are numerous ways to arrange secondary structures to achieve that general tertiary/quaternary shape.
I expect the second bullet to be more controversial than the first. Here is some documentation of the second point:
quote:
From structure to function: Approaches and limitations
Nature Structural Biology
November 2000 Volume 7 Number 11 pp 991 - 994
Janet M. Thornton1, 2, Annabel E. Todd1, Duncan Milburn1, Neera Borkakoti3 & Christine A. Orengo1
1. Biochemistry & Molecular Biology Dept., University College, Gower Street, London WC1 6BT , UK.
2. Crystallography Dept., Birkbeck College, Malet Street, London WC1 7HX, UK.
3. Roche Discovery Welwyn, Broadwater Rd , Welwyn Garden City, Hertfordshire, AL7 3AY, UK.
Correspondence should be addressed to J M Thornton. e-mail: thornton@biochem.ucl.ac.uk
Relationship between protein structure and biochemical function
There are now almost 12,000 protein structures in the Protein Data Bank (PDB)3, comprising over 25,000 domains. All these data support the hypothesis that there is a limited number of protein families/folds, despite the requirement for a vast array of different functions. Of the 2,159 structures determined in 1999, only 8% adopted novel folds4, 5. What have these revealed about the relationship between protein function and structure? The same fold is often seen in apparently different homologous families with different functions. (Fig. 2a). Martin et al.6 found little correlation between specific enzyme function and overall protein fold, in agreement with the observation that different folds can perform the same function, sometimes with the same catalytic cluster and mechanism (for example, trypsin and subtilisin). Indeed Hegyi and Gerstein7, using genome data, found two enzyme functions associated with seven different folds each. In contrast, several recent papers8 have highlighted the variety of biochemical functions that can be performed by proteins with the same fold or even by members of a single homologous family (Fig. 2b). Even within the PDB, with its relatively limited coverage, we have found 74 enzyme families with multiple functions8. Just as some folds occur very frequently9, we also find some families that are functionally highly promiscuous. Such families perform a vast array of different biochemical reactions, though these are often related in some way (for example one pyridoxal-phosphate (PLP)-dependent enzyme family is assigned 47 different Enzyme Commission (EC) numbers, which define the enzyme reaction catalyzed; Fig. 2c). In contrast, other families seem to confine their activities to a single function, although this impression may change with additional data. These results highlight the need to look beyond simple evolutionary relationships, at the details of a molecule's active site, to assign a specific function.
In other words, the same fold can often perform many different functions, and the same function can often be performed by many different folds.
Here is another example:
quote:
Genome Res 1998 Aug;8(8):779-90
Analogous enzymes: independent inventions in enzyme evolution.
Galperin MY, Walker DR, Koonin EV.
National Center for Biotechnology Information (NCBI), National Library of Medicine, National Institutes of Health, Bethesda, Maryland 20894, USA.
It is known that the same reaction may be catalyzed by structurally unrelated enzymes. We performed a systematic search for such analogous (as opposed to homologous) enzymes by evaluating sequence conservation among enzymes with the same enzyme classification (EC) number using sensitive, iterative sequence database search methods. Enzymes without detectable sequence similarity to each other were found for 105 EC numbers (a total of 243 distinct proteins). In 34 cases, independent evolutionary origin of the suspected analogous enzymes was corroborated by showing that they possess different structural folds. Analogous enzymes were found in each class of enzymes, but their overall distribution on the map of biochemical pathways is patchy, suggesting multiple events of gene transfer and selective loss in evolution, rather than acquisition of entire pathways catalyzed by a set of unrelated enzymes. Recruitment of enzymes that catalyze a similar but distinct reaction seems to be a major scenario for the evolution of analogous enzymes, which should be taken into account for functional annotation of genomes. For many analogous enzymes, the bacterial form of the enzyme is different from the eukaryotic one; such enzymes may be promising targets for the development of new antibacterial drugs.
...which is usefully free online here at www.genome.org.
Here is a quote from the conclusion:
quote:
Analogous Enzymes and Enzyme Classification
We showed that numerous biochemical reactions may be catalyzed by enzymes without detectable sequence similarity and, in some cases, with demonstrably distinct 3D structures. In other words, many enzymatic activities have been independently invented in evolution on more than one occasion. This is not to say that these enzymes have nothing in commonalthough not sharing common ancestry, they still may have similar reaction mechanisms and even similar local active center geometries. Identification of such common features in analogous enzymes seems to be an interesting direction for future research that may shed new light on mechanisms of enzymatic catalysis.
The current classification of enzymes on the basis of the catalyzed reactions is an indispensable tool for enzymologists, but in the case of analogous enzymes, it inevitably fails. A hierarchical system of protein classification constructed by sequence and structure comparison, like the ones already proposed for peptidases (Barrett 1994) and glycosidases (Henrissat and Davies 1997), can handle these cases and will provide a wealth of information complementary to the information currently embodied in the EC system.
They give the (long) list of their examples here:
Table 1. Dissimilar enzymes catalyzing same biochemical reactions
http://www.ncbi.nlm.nih.gov/Complete_Genomes/analenzymes.htm
...although I suspect that, since they are comparing proteins from different domains, and often the structures have not been determined yet, that some of these putative analogy cases will be determined to be interdomain homologs on the basis of structural similarity (the case for which I will make, eventually).
(also notable is the right column, where the same fold is used for different functions; the authors note that these are probably all examples of gene recruitment, where new genes are created by the cooption of old genes with different functions)
Some more articles with various degrees of relevance to the Genome Research article I quoted are listed here (although they do show just how much literature there is on topics like the evolution of new enzymes with new functions, the study of sequence and structural divergence, etc.).
Well, it took me too long to look through all of these articles so I won't be able to finish properly for now.
Conclusions: (what I would have got around to saying in more detail)
1) Both diverse sequences and diverse structures can generally carry out the same function.
2) Therefore, not only is sequence similarity between proteins a good indicator of homology, but structural similarity is also (or more accurately, these similarities are homologies in Owen's sense).
3) These similarities, being homologies, are not explainable by reference to a common function, either via a common designer designing for the same function, or natural selection designing for the same function.
4) These similarities can be explained by common descent, or possibly by an "archetype" idea, however the archetype/common plan explanation is wanting as there is no obvious reason why a divine/superadvanced designer should employ an archetype, which (by Owen's reasoning, which I've been following) is extra similarity not necessary to achieve the similarity of function.
5) If I had time I would use the above reasoning to argue that
(a) most homology inferences, e.g. where proteins are above the "twilight zone" of amino acid similarity (>25-30%), are completely noncontroversial and are not at all fuzzy/question begging/etc., as implied by Jay & Mike Gene
(b) given the references I provided regarding structural diversity for similar functions, similar protein structure is also a perfectly valid indicator of homology (although not quite as strong as structure + sequence), and that on this basis we can reasonably classify even such cases as ftsZ-tubulin and the DNA sliding clamp proteins as homologous, as scientists do (leaving aside the question of whether or not homology indicates common descent for the moment).
yersinia
PS: Following up one of Jay's links I found this very intriguing graphic which indicates that all of his talk about the drastic differences between different DNA clamp proteins is for naught:
Here is the link:
http://www.arn.org/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=1;t=001728
And the graphic I am thinking of:
Note how all of these rings, whether they have 2 parts (each 1/2 of the ring) or 3 parts (each 1/3 of the ring), have a sixfold structure. Why should the 2-part rings have 3 repeated domains per part, whereas the 3-part rings have 2 repeated domains per part? This seems like a very arbitrary similarity for a designer to stick in. However, this appears to be parsimoniously explained by postulating a common ancestor that made a DNA sliding clamp out of 6 1-domain parts (the whole ring could be encoded by one gene originally), and that in different lineages this gene was tandemly duplicated and merged into one gene with either two domains or three domains, respectively. It's a simple, elegant, explanation, and it actually explains, rather than explains away, the 6-domain structure shared by all of the diverse DNA sliding clamps.
IP: Logged
|
|
Jay
Member
Member # 268
|
posted 11. July 2002 02:55
Hi Yersinia,
Thanks for the replies! I'll just address one specific point here, and try to get to the rest tomorrow.
Yersinia: "Note how all of these rings, whether they have 2 parts (each 1/2 of the ring) or 3 parts (each 1/3 of the ring), have a sixfold structure. Why should the 2-part rings have 3 repeated domains per part, whereas the 3-part rings have 2 repeated domains per part? This seems like a very arbitrary similarity for a designer to stick in. However, this appears to be parsimoniously explained by postulating a common ancestor that made a DNA sliding clamp out of 6 1-domain parts (the whole ring could be encoded by one gene originally), and that in different lineages this gene was tandemly duplicated and merged into one gene with either two domains or three domains, respectively."
I was hoping we'd get to pick this one back up, as we seemed to let it go back at ARN.
Here's the basic problem with this scenario - the composition of the subunits themselves is wildly different - there is no suggestion from the actual makeup of these subunits that they ever were common 6-1 domains. It's a nice story, but it strains greatly after a closer look.
If you take alignments from the archaea and eukaryotic clamps, you can see that they stack nicely and give a clear protein type, with lots of distinct areas of high conservation. But then, if you move over to the beta clamps, you see a whole new story. The conserved regions within the beta clamp subunits are robustly very different from the PCNA (arch/euk) type, but are also conserved within their type.
Basically, there is no reason from the alignment data (which spans a very considerable amount of diversity in organisms, IMO) to ever assume that they came from common subunit building blocks. In fact, the data robustly argues against this.
However, you rightly point out that they are nontheless quite similar in their six-fold makeup, and in their overall shape. But that is precisely why I cite this example. They are *very* similar in function and in overall shape, but very, very different in actual composition. It is almost a contradiction in the data, from a naturalistic standpoint.
Most people, after looking at the crystal structures, would agree that these rings are related - and this usually translates into homology. I also argue that they are related, but for a much different reason - I argue that they share a common engineering idea. And from the data, I personally consider this a better interpretation than homology, 6-1 subunits notwithstanding.
Thanks,
jay
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 11. July 2002 05:31
quote:
Here's the basic problem with this scenario - the composition of the subunits themselves is wildly different - there is no suggestion from the actual makeup of these subunits that they ever were common 6-1 domains. It's a nice story, but it strains greatly after a closer look.
Please be specific by what you mean by "the composition of the subunits themselves". Be specific if you can -- are we talking no sequence similarity or what? What are the percentage similarities for various comparisons, e.g. between eukaryotes and archaebacteria, and between them and eubacteria. You've been making a lot of vague statements but I haven't seen much actual data.
I've been assuming that the sequence similarity was no better than chance and that all we had to go on here was secondary structure (you can see this similarity in the diagrams). But then you say this:
quote:
If you take alignments from the archaea and eukaryotic clamps, you can see that they stack nicely and give a clear protein type, with lots of distinct areas of high conservation. But then, if you move over to the beta clamps, you see a whole new story. The conserved regions within the beta clamp subunits are robustly very different from the PCNA (arch/euk) type, but are also conserved within their type.
If you're talking secondary structure alignments, I just don't see this huge difference that you're talking about. Looking at the diagram, above, looks like the PCNA & beta clamps have massive structural overlap.
Are you saying that the arch/euk clamps are more similar to each other (sequence or structure, its hard to tell what you mean) than to bacterial clamps? If so, this would fit in quite nicely with a similar pattern widely established for core informational genes (and supporting the hypothesis that eukaryotes evolved from archaebacteria, with a large infusion of metabolic genes from a eubacterial symbiont). This is not a very radical finding if so, rather it conforms with expectations.
yersinia
PS: What do you mean by "protein type"? Are you just talking about the difference in subunit size again? It appears that the 6-subunit common ancestor hypothesis offers an explanation for this, I haven't seen another explanation offered yet other than a vague appeal to common design.
What's so hard to believe about a gene being duplicated, and then having the separator between them deleted, making a new gene of twice the size (or repeat with one more gene for a 3x gene). A few billion years of basically neutral divergence (same function being maintained) randomizes the sequence similarity, and voila. Only common mutational mechanisms are invoked, and only a few of those. It looks like a quite elegant explanation from here, but you're not giving it the time of day.
PPS: This specific example is just an aside from the general argument I made the previous that homology is a good inference even if its only based on structural data.
[ 11 July 2002, 05:34: Message edited by: yersinia ]
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 11. July 2002 06:02
Very briefly,
I just came across this article which indicates that we might not have gotten the full story even with the sequence similarity of sliding clamps just yet in the thread:
(adding the emphasis)
quote:
To reach a clearer understanding of the origin(s) of the DNA replication system by comparative analysis of the sequences and structures of their components, additional, systematic effort in two directions seems to be necessary: (i) detecting subtle sequence and structural similarities that have escaped detection previously; (ii) solving the issue of orthologous relationships between replication components. The importance of the former aspect is underscored by the homologous relationship between the bacterial and eukaryotic sliding clamp proteins that was not originally recognized but became apparent when their structures had been determined (7,27). With the advent of more powerful methods for sequence analysis, such as PSI-BLAST (28), the similarity between the clamp proteins has become detectable at the sequence level. This suggests that systematic, careful comparisons of replication proteins might reveal additional subtle but evolutionarily and functionally important similarities. Such findings could shift the balance in our thinking about the evolution of DNA replication towards the common origin hypothesis, whereas the absence of detectable similarity in spite of a careful comparison might suggest independent origin for at least some of the components. It is critical for any meaningful evolutionary reconstruction to distinguish orthologs that likely evolved from an ancestral component of the replication machinery from homologous but not orthologous proteins that might have independently originated from proteins that had functions other than DNA replication.
Here is the article:
Did DNA replication evolve twice independently?
http://nar.oupjournals.org/cgi/content/full/27/17/3389
Oops, there's a little more:
(emphasizing the bit on sequence as the structural similarity appears to be agreed upon by everyone)
quote:
Other proteins and domains, namely archaeal/eukaryotic FEN1/RAD2 nucleases and bacterial 5[prime]->3[prime] exonuclease domains of polymerase I, the replication sliding clamps (PCNA) and DNA ligases (the NAD-dependent ligase in bacteria and the ATP-dependent ligase in eukaryotes), show very low sequence conservation but, nevertheless, appear to be orthologs (Table 2 and Fig. 1). Until recently, the homologous relationships between these components of the replication machinery remained undetected. However, detailed sequence comparisons as well as structural superposition for the sliding clamps and the ligases (36,52; see also above) indicated that in each of these cases, the bacterial and archaeal/eukaryotic proteins are homologous. Moreover, apparent horizontal gene transfers apart, the bacterial proteins in each of these cases are more similar to their functional counterparts from archaea/eukaryotes than to any other archaeal or eukaryotic proteins (Table 2). These observations suggest that orthologous relationships exist for each of these proteins, in spite of the high level of divergence.
So perhaps even some sequence similarity exists for the DNA sliding clamps.
yersinia
IP: Logged
|
|
James A. Barham
Member
Member # 50
|
posted 11. July 2002 11:07
Yersinia:
Thank you for the extremely lucid postings, and for the very useful references, as well.
My own interest is somewhat tangential to the main topic of this thread, but I thought I would take advantage of your intervention to ask your opinion on something.
I do not question common descent, although I am aware of the difficulties in establishing particular lineages with anything like certainty. But I accept that in general your (or rather Richard Owen's) distinction between analogy (common function with different structure) and homology (common structure with different function) is real and important.
My main interest relates, rather, to the general explanatory adequacy of selection theory and a general reductionist attitude toward life to explain the phenomena of evolution. I don't want to belabor points again here that I have raised repeatedly elsewhere on Brainstorms (about the way in which both molecular biology and selection theory presuppose the normative functional integrity of the cell, and so cannot explain the source of normativity in nature).
What I would like to focus on here, instead, is the question of archetypes as an explanation for analogies, as opposed to "selection pressure" as the main mode of explanation of these striking phenomena. You only allude to archetypes' being in the mind of God, but what if they are somehow immanent in the dynamics of the cell itself? More specifically, I would be very interested to know what you think of recent (and not so recent) "structuralist" trends in biological thought?
I am thinking, for example, of D'Arcy Thompson's "On Growth and Form" (orig. publ. 1917, numerous editions), of Gerry Webster and Brian Goodwin's "Form and Transformation" (Cambridge UP, 1996), of the numerous Santa Fe Institute and A-Life studies on the abstract patterns underlying biological order, and the fewer but very interesting efforts to specify an abstract "morphospace" for a particular type of organic form quantitatively (e.g., R.D.K. Thomas and W.E. Reif, "The Skeleton Space: A Finite Set of Organic Designs," Evolution, 1993, 47: 341--360).
I know that most Darwinian thinkers are fairly comfortable with this work, feeling that it will simply be incorporated into an expanded Darwinian "paradigm." I personally feel it is far more subversive than that. But, at any rate, I was just wondering what you thought about the naturalistic, structuralist alternative to understanding analogies as archetypes.
[ 11 July 2002, 11:45: Message edited by: James A. Barham ]
IP: Logged
|
|
|
Printer-friendly view of this topic