|
Author
|
Topic: Denton on Protein Folds
|
Frances
Member
Member # 169
|
posted 11. March 2003 11:19
Mike,
I think that the answer may be quite simple, homology as Yersinia also points out is not just infered from structure similarity. See my reference for instance.
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 11. March 2003 12:01
Yersinia:
*******************
Yersinia wrote:
Finally, we should keep in mind that the structural homology inference is supported by the following (IMO fairly compelling) pattern found in many proteins:
100% sequence similarity -- superimposable structures
90% sequence similarity -- superimposable structures
.
.
.
20% sequence similarity -- superimposable structures
(approaching random sequence similarity)
10% sequence similarity -- superimposable structures
...and this decay in sequence similarity follows the usual patterns of closeness-of-lineage-relationship.
*******************
Consider the set of all sequences that code for the globin fold. Now make a scatter plot of these sequences in sequence space. The unspoken assumption in evolutionary theory seems to be that this scatter plot forms a single, large, contiguous set. That is, using single-point replacements one can move anywhere within the set. The set is fully traversable via single-point replacements. The evolutionary idea being that as evolution occurs, the fitness landscape across this set is more or less flat and so the globin randomly evolves, being restricted only by the set boundaries. IOW, a protein sequence is determined mostly by the contingencies of history as opposed to any functional or design necessity.
Aside from this hard over "contingency dominates necessity" assumption, this assumption that the set is traversable via single-point replacements is unsubstantiated and in fact problematic. Problematic because the set would have to span a good portion of sequence space since it includes very different sequences with practically no similarity (indeed, one can find examples at or even below the random threshold). It is difficult to see how the few thousand sets, for all the different protein folds, would not be massively overlapping in sequence space. Also, we know of relatively similar sequences (50% or greater %ID) which do not assume the same fold). Hence, we must at the very least go to a thin-bridge model, where the very disparate sequence clusters found within a protein family such as the globins are connected only by thin bridges such that the set is now a tiny fraction of its convex hull. This way we now accommodate all the different protein folds.
But the thin bridge model is hardly parsimonious or natural. Why would such thin bridges exist, connecting such otherwise disparate sequences? Clearly, such a model is motivated by a prior commitment to evolution. Finally, we are now finding that the folding pathways are different for the disparate sequences within a family. So again, the more natural view is that we have separate clusters, coding in their own way, for the same fold.
Evolution's hard over "contingency dominates necessity" assumption, (vis-à-vis protein sequence – structure), is itself problematic because, while it is true that structure can be coded for by very different sequences, there are nonetheless many potential functions and reasons for an amino acid selection at a particular loci in the sequence. This assumption relies on the rather outdated single-sequence, single-molecule, single-pathway view of the cell. IOW, the view that these entities operate more in isolation than we are now finding. The cell is more holistic than this, and structure is only one effect of the amino acid sequence.
But more to the point here, the "fully-traversable set" assumption is also problematic for the reasons stated above. Nonetheless, evolutionists point to the protein sequence – structure data as strong and compelling evidence for their theory.
I see this as an example of subtle circular reasoning. IOW, this "fully-traversable set" assumption is based on the presupposition that evolution is true and this assumption is required to interpret the data so it serves as strong and compelling evidence for evolution.
--Cornelius
IP: Logged
|
|
Mike Gene
Member
Member # 149
|
posted 11. March 2003 13:03
Frances: I think that the answer may be quite simple, homology as Yersinia also points out is not just infered from structure similarity. See my reference for instance.
Sure, but in many cases, people are relying primarily on structural similarity to declare homology. These are the instances that get called into question.
Let me say something about the arc repressor protein (something you described as a "beautiful examples of protein evolution").
By simply interchanging N11 and L12, the researchers were able to bring about a dramatic change in protein structure shown below:
What is beautiful about this is how form dominates. That is, the beautiful form seen in both the WT and mutant strain is equivalent, yet the form in the mutant strain is not a consequence of natural selection fine-tuning anything. As Denton would argue, natural selection might be able to exploit this form, but the mutant form is primary. All of this speaks to life being designed to make smart use of natural selection.
However, we need to keep in mind that this mutant was created with cassette mutagenesis to bring about a simultaneous swap of these residues. Furthermore, if you consult the genetic code, asn and leu cannot be bridged with a single base substitution. One could, through single steps, approach the swap. For example, we could move from leu -gln - asn. And from asn - ile - leu. Nevertheless, these would not be simultaneous.
Now, I am not arguing for any insurmountable barrier between these sequences/forms. But it does raise the tantalizing hypothesis that the genetic code might mediate transitions between forms, such that some transitions are much more likely than others. Given that I have already provided evidence that the genetic code can be used to channel the effects of common base substitutions, this would not be surprising.
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 11. March 2003 18:55
Yersinia, what you say about the descending sequence of similarity with superimposable structures is true, but that doesn't mean that the superimposable structures necessarily tell you very much at the 10% amino acid sequence similarity level. (Especially if it's only similarity, not identity!)
You might be able to use lineage patterns, slowly evolving genes, and whatnot, to infer that two organisms are related, and then the most parsimonious explanation for superimposable structures might be that they were related (if the divergence time was long enough such that you should be down to 10% similarity). But the strength of the parsimony depends on how easy it would be to get the same structure via a different origin, and if it is thought to be fairly easy, I'm not sure you could conclude much from superimposability. And you certainly couldn't conclude much about the existence of a common ancestor from superimposability, in that case.
Cornelius, I would agree that the amino acids distribution found in the globin-fold is weirdly clumpy, but keep in mind that the extant organisms share common ancestors to weirdly clumpy degrees, and that is where we get the globin folds from. So it is not enough to simply note that there are thin bridges, and that this is unnatural--because the pattern of extant life on the planet makes it very hard to tell whether the bridges are indeed thin in terms of amino-acid-sequence space. They may be very broad bridges, but extinction may have left only islands of similarity remaining.
The distinction ought to be testable by comparing multiple folds and seeing if isolation occurs at similar levels, but there are probably additional considerations that I haven't thought of yet.
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 11. March 2003 19:16
Rex:
Good points. I agree that the thinness or thickness of potential bridges is difficult to assess. Basically, we're dealing with a high-dimensional space with only a few data points sprinkled about. I suspect that if we were able to see the actual topology of the set of all sequences that produce the globin fold, that it wouldn't look very pretty.
But keep in mind that when you say "the extant organisms share common ancestors to weirdly clumpy degrees, and that is where we get the globin folds from" you are presupposing evolution; something that I am not doing.
--Cornelius
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 11. March 2003 21:15
I was speaking from an evolutionary perspective, which you must do also to claim that the data is poorly accounted for by evolution.
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 11. March 2003 22:24
Rex:
You wrote: "I was speaking from an evolutionary perspective, which you must do also to claim that the data is poorly accounted for by evolution."
Ah, OK, I see. Then let me clarify just a bit that I'm not saying that the data are evidence *against* evolution so much as they are not strong evidence *for* evolution. I don't think we know enough to say very confidently. What intrigues me is the rather strong claims that have traditionally been made in evolution's favor regarding these data.
--Cornelius
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 12. March 2003 00:14
Agreed. Structural homology doesn't appear to be very strong evidence at this point.
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 12. March 2003 12:28
Mike,
I believe that we all would agree that single base mutations are not the only forms of genetic variation.
I am still looking for ways to identify the causal direction, was life designed to make smart use of natural selection or was natural selection that made life appear to be designed smartly? So far the evidence at most shows that life is opportunistic and in fact applying Dembski's methodology for inferring design, since chance/regularity cannot be eliminated, ID cannot be inferred. What methodology do you propose to resolve this? Or can it even be resolved? You suggest in another thread that front loading does not require 'supernaturalism' but in this thread you seem to suggest that the laws of nature which guide the protein (fold) distributions and amino acid distributions have been manipulated in some form or manner.
What, when, how and where are some of the questions that come to mind.
IP: Logged
|
|
Mike Gene
Member
Member # 149
|
posted 12. March 2003 14:19
Frances: I am still looking for ways to identify the causal direction, was life designed to make smart use of natural selection or was natural selection that made life appear to be designed smartly?
A classic example of looking at the same thing and being able to interpret it differently. I choose to explore the first interpretation for several reasons. Most important is the fact that it is the perspective that is essentially ignored and thus offers tantalizing promises. And you have offered nothing that would lead me to abandon this first alternative.
Besides, natural selection is a blind jury-rigger and we're talking about the matrix that the blind jury-rigger needs to function. In this case, it doesn't seem to be natural selection that is crafting these folds.
So far the evidence at most shows that life is opportunistic and in fact applying Dembski's methodology for inferring design, since chance/regularity cannot be eliminated, ID cannot be inferred.
I infer design because many features of life exhibit the properties we would expect to exist if design were true. As for opportunism, it is becoming more and more clear (to me) that the life was designed to exploit opportunism. Shapiro's 'natural genetic engineering' thesis is a great example of this.
What methodology do you propose to resolve this? Or can it even be resolved?
I addressed this before.
quote:
You want ways to distinguish between front loading design and natural processes. Yet it's way too early in the investigation to provide such a definitive answer. At this stage in the game, FLE represents an alternative perspective. That is, if the two perspectives cannot be distinguished, there is no reason for me to ignore the FLE perspective and focus entirely on natural processes to the exclusion of their possible teleological use. Now, after the FLE perspective has been thoroughly fleshed out, to answer several of the questions I have posed, we can then return to your question. That is, let's first see what falls out from a more focused FLE perspective, okay?
You suggest in another thread that front loading does not require 'supernaturalism' but in this thread you seem to suggest that the laws of nature which guide the protein (fold) distributions and amino acid distributions have been manipulated in some form or manner.
The laws are the laws. The point is that any intelligent designer would seek to exploit the laws to further their design objectives.
What, when, how and where are some of the questions that come to mind.
What? The origin of the first cells on this planet.
When? Approximately 3-4 billion years ago.
How? Advanced biotechnology that may also rigged evolution through its implementation.
Where? On earth.
Sorry Frances, but it seems every time I raise a new, focused point, you try to steer the discussion into generic philosophical issues. I've already answered your generic front loading questions. Whether the answers personally satisfy you are not important. Let's get back to the topic of this thread.
IP: Logged
|
|
Moderator
Administrator
Member # 1
|
posted 12. March 2003 15:38
Frances: Persist and be restricted. This discussion between you and Mike has been had before and it does nothing but create noise in an otherwise fun thread.
IP: Logged
|
|
Mike Gene
Member
Member # 149
|
posted 12. March 2003 23:20
I thought I would remind people that Denton is supposed to chat about this paper next month.
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 12. March 2003 23:49
Hi Mike
Are you by any chance familiar with Newman and Muller's work? They seem to have come to very similar conclusions as yours. Their scientific work is indeed fascinating to read
quote:
The earliest epigenetic mechanisms to influence biological form were the physics of chemically active condensed materials, which include primitive cell masses, resulting in a delimited, and essentially exhaustive, array of body plans and organ forms—segmented, hollow, multilayered, and branched structures.
As a consequence of the biochemical and genetic integration of interactions, development increasingly takes place in a Mendelian arena in which genotype and morphological phenotype become more closely matched. Development also becomes susceptible to Darwinian modification leading to the exploration of the residual morphogenetic “play” remaining in multicellular systems. In particular, physical properties and threshold effects of the developmental systems under modification generate morphogenetic by-products that become the kernels of morphological innovations, which elaborate on a smaller scale the major morphological themes of the earlier phase.
I will discuss this thread in more detail in my RNA thread. It seems that like RNA, protein networks seem to be very similar.
Fascinating
A late add
quote:
We emphasize that the indirect relationship of genes to form, which we postulate for tissue morphogenesis, is analogous to what is generally accepted to constitute this relationship in the most fundamental role of genes: protein synthesis. Here genes also influence the realization of form without being its determinants. The three dimensional, folded structure of a protein—its biologically functional
morphology—is defined by interactions of the polypeptide chain within itself and with its external environment. The typical functional form of a protein is identical to that decreed by the thermodynamics of spontaneous processes. Correspondingly, the universe of protein secondary structures and folded motifs in existing organisms is limited to a relatively small number of forms perhaps 1,000) out of an astronomically large number of potential random compact structures Chothia ’92; Li et al. ’96).
and
quote:
Just as an understanding of the set of preferred protein motifs and the morphologies of particular proteins depends on an appreciation of the originating role of physical mechanisms, we contend that an understanding of the forms assumed by metazoan organisms requires knowledge of the generative epigenetic processes that originally (in evolutionary history) produced those forms.
[ 13. March 2003, 01:55: Message edited by: Frances ]
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 13. March 2003 00:41
Sorry I dropped this thread, I wrote half of a more detailed reply on homology but then paused and haven't gotten back to it.
One important question:
- There are clearly examples where dwindling sequence similarity can be traced right back to nothing and yet structure remains the same and no one (AFAIK) questions common descent of the proteins. E.g., some MotAB proteins are so far diverged from e.g. the E. coli MotAB protein that the e-values measuring similarity are 0.05 or higher. Either we have to postulate independent origins for the motor of the flagellum, or these sequences have diverged to random yet the structure has remained. I suspect that there are a lot of documented cases like this.
- On the other hand, are there any clear cases of the independent origin of the same protein structure?
There certainly are examples of functional convergence (e.g. archaeal vs. bacteria flagellum, or this list of examples in enzymes:
Table 1. Dissimilar enzymes catalyzing same biochemical reactions
http://www.ncbi.nlm.nih.gov/Complete_Genomes/analenzymes.htm
And in general I would recommend the pretty-good Convergence and Divergence thread from last year to e.g. Rex who probably arrived after it was over since I don't have time to re-spout all my views on the definition of homology) -- but tellingly the functional convergence does not seem to lead to structural convergence.
nic
PS: While the discussion continues I would also like to re-emphasize that there is more to a protein than a fold, so even if one were to concede completely to Denton's argument about folds there would still be much structural similarity that would have to be accounted for.
PPS: For Cornelius: surely the "bridge" model is supported by more than mere assumption of common descent! For almost any protein you can stick it in a search and get a gradually descending sequence similarity that correlates with decreasing relationship in traditional taxonomy. Usually the differences are just a few percent or one or a few amino acids at each "step", and data is pretty coarse at the moment as we only have a few metazoan genomes done now. Perhaps we can't quite document a continuous bridge yet but there are a great many observed stepping stones close to each other.
Plus, it is well known that many amino acid changes are fairly minor anyhow, e.g. switching one nonpolar side chain for another.
Finally, natural selection is just the kind of process that can keep a population on a narrow bridge -- the individuals with the nonfunctional mutations will fall off the bridge, but as a result these variants never spread and only the successful variants can persist.
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 13. March 2003 02:21
Yersinia:
************************
You wrote:
PPS: For Cornelius: surely the "bridge" model is supported by more than mere assumption of common descent! For almost any protein you can stick it in a search and get a gradually descending sequence similarity that correlates with decreasing relationship in traditional taxonomy. Usually the differences are just a few percent or one or a few amino acids at each "step", and data is pretty coarse at the moment as we only have a few metazoan genomes done now. Perhaps we can't quite document a continuous bridge yet but there are a great many observed stepping stones close to each other.
Plus, it is well known that many amino acid changes are fairly minor anyhow, e.g. switching one nonpolar side chain for another.
Finally, natural selection is just the kind of process that can keep a population on a narrow bridge -- the individuals with the nonfunctional mutations will fall off the bridge, but as a result these variants never spread and only the successful variants can persist.
*******************
Yes, I agree with what you are saying. Perhaps I made the point about the bridge model too strongly. But I wouldn't go too far either. For example, how big are the "islands" which may or may not be connected? Maybe an island contains sequences as divergent as 30% ID or less.
What we do know is that (i) very different (randomly related) sequences can code for the same fold and (ii) a given sequence has many close neighbors (say 50% ID or greater just off hand) which presumably are on the same island.
Until we get examples of bridges consisting of closely spaced sequences (only a few % different at each step) leading all the way from two distant islands, then it is not clear to me why we should assume bridges exist, unless we're presupposing common descent. But this is not to say that there isn't, as you point out, at least the suggestion of bridges from the available data.
You write: "For almost any protein you can stick it in a search and get a gradually descending sequence similarity that correlates with decreasing relationship in traditional taxonomy." Agreed, I believe I said as much. But again, this can be explained from a design perspective if one if willing to forego the "contingency dominates necessity" assumption and reckon that there could be (non structural) functional reasons for amino acid assignments.
Regarding the idea that NS is just the tool for keeping folks on the bridges. Yes, again I agree, but it becomes a question of kinetics. IOW, how thin and how long is the bridge? One could argue there is a bridge between lead and gold, or between a protein native conformation and a certain misfold. But you never or rarely see them convert because the time constant is billions of years or more. I don't intend these as perfect analogies.
My point is simply that the existence of a bridge does not alone solve all problems. I would argue that fish did not evolve into amphibia, for example, but I'm not sure there is *no* viable bridge of closely-spaced intermediates between them.
--Cornelius
[ 13. March 2003, 02:23: Message edited by: Cornelius G. Hunter ]
IP: Logged
|
|
|
Printer-friendly view of this topic