|
Author
|
Topic: Reverse Engineering Evolution; evolvability and the primitives responsible for life
|
Jack Foster
Member
Member # 79
|
posted 24. August 2003 13:32
Reverse Engineering Evolution; evolvability and the primitives responsible for life and consciousness
1. Evolvability
Daniel Dennet once wrote:
“…evolution will occur whenever and wherever three conditions are met: replication, variation (mutation), and differential fitness (competition)”.
Dennet thought that replication, variation, and selection are sufficient to provide evolution, but experiments in artifical evolution have proven Dennet wrong. There is indeed a fourth condition: evolvability.
From Wagner and Altenberg:
"In evolutionary computer science it was found that the Darwinian process of mutation, recombination and selection is not universally effective in improving complex systems like computer programs or chip designs. For adaptation to occur, these systems must possess "evolvability". . ."
To say that evolvability is a condition for evolution is circular if we think of evolvability only as "that which provides evolution". Clearly we need to understand evolvability. There are many definitions of evolvability:
"the capacity to evolve."
"the ability of random variations to sometimes produce improvement."
"the ability of a population to produce variants fitter than any yet existing."
"the capacity for an evolutionary system to generate adaptation."
And here's an interesting "definition" of sorts from Peter Turney:
"It is difficult to define evolvability, beyond saying that it is the capacity to evolve. We suggest the following sufficient (but not necessary) condition for evolvability: If individuals A and B are equally fit but the fittest child of A is likely to be more fit than the fittest child of B, then A is more evolvable than B."
I'm going to add one more definition which provides another way to think about the notion. Evolvability is the selection-friendly relationship between genotype phase space and fitness phase space. When something is evolvable, it has a rich, multi-dimensional fitness phase space, and when something is not evolvable, it's fitness space is dimension poor. For the unevolvable, there may be an available mutation that increases fitness in some fitness plane, but selection has such a limited fitness space upon which to work, that once the few beneficial mutations are found in limited space, evolution is finished.
Fitness space is internally-derived; it is a property of the replicator itself - a function of the variability of the replicator. Fitness landscapes upon that space are externally determined. If we think of computer programs as genotypes, we can think of the operation of the programs as phenotypical. You can then imagine the operation of a program as occupying a unique position within phenotype space, and imagine that each phenotype has its own (though perhaps shared) fitness phase space. Since functional programs as traditionally written are intolerant to change, they can be thought of as occupying "fitness space islands" since there's no way to acquire positive adaptations by randomly altering the program. Variations will typically randomize or "break" the program, therefore fitness phase space is without dimension, or at least is dimension-poor.
The way to attain evolvability with computer programs is by the careful design of the data structure called the genotype to phenotype map. (g-p map) A carefully designed g-p map constrains phenotype space in a way that is friendly to evolution; it provides reach across vast (formerly) unfriendly landscape. Genes may be mapped only to functional phenes, so fitness space islands come together to form something larger! Dimensions are added to fitness space.
2. Reverse Engineering
It is often argued that ID is an argument from ignorance, and that to throw up hands and declare "God did it" does not provide a perspective with potential for discovery. I think that the useful perspective that ID naturally leads to is one of reverse engineering. How was evolution accomplished? This is the perspective that Mike Gene assumes, for instance.
Strict Darwinists imagine Natural Selection as all-powerful. But now that we have experience with thousands of designed evolutionary systems, we know that each system has a unique potential evolutionary trajectory independent of selection. Tierra evolves differently than does Sims' Virtual Creatures than does various AVIDA runs. Evolvability lives on the variation side of the ledger, NOT on the selection side. What primitives are necessary or sufficient to evolve life's organisms? Can we reverse engineer evolution and evolvability to better understand the success of biological evolution? Certainly the potential for that success was inherent in life's initial condition. So what about the initial condition allowed for such vast potential?
[ 24. August 2003, 13:58: Message edited by: Jack Foster ]
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 24. August 2003 14:45
Jack: It is often argued that ID is an argument from ignorance, and that to throw up hands and declare "God did it" does not provide a perspective with potential for discovery. I think that the useful perspective that ID naturally leads to is one of reverse engineering.
Reverse engineering systems does not require ID as a premise. In fact I fail to see how ID adds anything new to how science goes about its business.
Evolvability is indeed an interesting topic and I would suggest a possible answer "neutral networks"
The work of Schuster, Fontana and many others in this area may be helpful
Ancel and Fontana "Plasticity, evolvability and modularity in RNA"
quote:
"Evolutionary search has two stages: neutral walk and fitness improvement. During neutral walk, both active and inactive genes change.Hypothesis: A satisfactory adaptive/neutral mutations ratio
during neutral walk can lead to fitness improvement."
Conclusions:
There is a positive relationship between neutrality and evolvability in a Boolean function landscape.
When the adaptive/neutral mutations provide a satisfactory exploitation/exploration ratio,
evolution is able to generate fitter offspring.
From: Neutrality and the Evolvability of
Boolean Function Landscape, Tina Yu and Julian Miller
"The Evolution of Evolvability in Genetic Programming" Altenberg
quote:
Evolutionary algorithms apply the process of variation, reproduction and selection to look for an individual capable of solving the task at hand. In order to improve the evolvability of a population we propose to copy important characteristics of nature's search space. Desired characteristics for a genotype-phenotype mapping are described and several highly redundant genotype-phenotype mappings are analyzed in the context of a population based search. We show that evolvability is influenced by the existence of neutral networks in genotype space. The extent of the neutral networks affects the interconnectivity of the search space and thereby affects evolvability. Species evolving on a non-redundant mapping reach a state of stasis after a few number of generations. In effect, evolution comes to a halt. However, species evolving on a genotype-phenotype mapping with extensive neutral networks are continuously able to find adaptive mutations and are able to locate higher optima. The existence of highly intertwined neutral networks increases the evolvability of a population.
Marc Ebner, Patrick Langguth, Jürgen Albert, Mark Shackleton and Rob Shipman. On Neutral Networks and Evolvability
or Marc Ebner, Mark Shackleton and Rob Shipman.
How neutral networks influence evolvability.
Complexity, 7(2), pp. 19-33
From what I understand, the scale free network combined with neutral evolution is an essential component for evolvability.
In fact robustness, modularity, plasticity and evolvability seem to be all related.
See also Special Journal Issue of
BIOSYSTEMS Journal of Biological and Information Processing Sciences
[ 24. August 2003, 14:51: Message edited by: Pim van Meurs ]
IP: Logged
|
|
Jack Foster
Member
Member # 79
|
posted 24. August 2003 23:59
Your references are consistent with what I'm saying. Here from your last reference:
quote:
Until the 1990's biologists had generally been concerned about the fate of variation rather than its origin, tacitly assuming that the origin and maintenance of variability was an automatic by-product of Darwinian evolution. Experiments with computational or simple organic evolutionary systems show emphatically that this need not be the case, and the resulting explanatory gap now poses a serious challenge for evolutionary theory. Currently evolutionary biology can say little more than that these properties arose somehow in the course of organic evolution on earth, therefore interest in this topic is now very high.
Darwinian evolution, characterized by heritable variation and selection, is not by itself sufficient to account for the capacity to vary, the specific type of variability present, nor for heritability of phenotypic fitness. Rigidity of genotype-phenotype mappings, as often used in evolutionary computation or population genetics, constrains the dynamics of evolution to a small space of possible biological or artificial systems. Open-ended evolution is not possible under such constraints. Yet evolution, by itself, cannot fully explain the advent of genetic systems, flexible genotype-phenotype mappings, and heritable fitness. This presents a challenge both to biologists seeking to understand the capacity of life to evolve and to computer scientists who seek to harness biological-like robustness and openness in the evolution of artificial systems. The sources of variability and its transmission between generations have been identified as key to biological evolvability. Properties such as the facilitation of extradimensional bypass and robustness to genetic variability (Conrad, 1990), heritability of fitness (Michod & Roze, 1999), modularity, as well as robustness to developmental variation (Kirschner & Gerhart, 1998) play important roles in evolvability.(my bolds)
quote:
Pim: Reverse engineering systems does not require ID as a premise.
I believe that evolutionary biology's focus on selection did not provide encouragement for the reverse engineering of the evolutionary system called life. If Natural Selection is all-powerful, what's the point of focusing on variation? Yet that's precisely where design would occur and where evolvability lives. Evolvability apparently requires a gene to phene map, and therefore . . . non-random primitives.
I agree that Science sometimes borrows the perspective of reverse engineering, but reverse engineering is fundamentally a teleologic perspective. Reverse engineering implies engineering.
Dembski has argued inductively that design can be inferred from CSI because every example of CSI where we have first hand knowledge of origin was indeed designed. Critics (incorrectly I believe) respond that CSI simply means improbable. But let me make Dembski's same inductive argument with evolutionary systems. We have experience with thousands of evolutionary systems. Never has a system with evolvability popped up by chance; all require careful design.
So let's get the experts in here to design life. What type of non-random primitives would be necessary or sufficient to evolve what we see in life? What is it about those primitives that enable characteristics like consciousness and intelligence?
quote:
Pim:In fact I fail to see how ID adds anything new to how science goes about its business.
So why do you hang out here?! I thought this was "Brainstorms"! ![[Wink]](wink.gif)
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 25. August 2003 00:45
Jack, One might be careful to portray Darwinism, that is Darwin's ideas, to focus exclusively around natural selection.
Neutral 'evolution' needs hardly be a problem for Darwinian evolution since as you point out, it deals with increasing variation. Only at a later time does selection come into play.
The combination of scale free networks, and neutral evolution with selectable evolution make for a powerful mechanism. Whether or not it is a serious challenge for evolutionary theory as it presently stands (Neo-Darwinism for example), it surely should not be mistaken as evidence for intelligent design since I believe that the various research shows how the combination of various known processes can help us understand evolution, Darwinian evolution amd issues such as evolvability, modularity, degeneracy etc. The call for papers may be overstating the issue slightly as evidenced by the papers on this topic which seem to all point to the relevance of scale free networks and neutral evolution in addition to selectable evolution. Scale free networks seem to form a powerful witness to the mechanisms of evolution.
Jack: Dembski has argued inductively that design can be inferred from CSI because every example of CSI where we have first hand knowledge of origin was indeed designed. Critics (incorrectly I believe) respond that CSI simply means improbable. But let me make Dembski's same inductive argument with evolutionary systems. We have experience with thousands of evolutionary systems. Never has a system with evolvability popped up by chance; all require careful design.
CSI means improbability for the simple reason that specification is trivial and complexity is equivalent to probability. Your suggestion that evolvability requires 'careful design' seems to go against what I have read on these topics. In fact I would argue that evidence shows the contrary namely that careful 'design' is not required for evolvability to happen. One example would be hypermutability for example, in times of stress organisms can increase their mutation rates, searching for solutions to their stress. Others include the importance of neutral space. A great example is the evolution of silicon networks which were tasked to find a solution to distinguish between two tones. A beautiful case of neutral bridging seems to have taken place here.
"Through the labyrinth evolution finds a way: a silicon ridge" by Inman Harvey argues that neutral paths through genotype space may be essential for the effectiveness of evolution. And guess what RNA genotype space seems to be especially conducive to neutral evolution, scale free, closeness of various phenotypes etc. It may be that the mechanism of evolution, gene duplication and addition of function may be helpful in understanding the evolvability of evolution.
I would be interested though to hear what the ID perspective on this issue would be?
An interesting paper
"Neutral and nonneutral mutations: the creative mix--evolution of complexity in gene interaction systems. Zuckerkandl E. J Mol Evol. 1997;44 Suppl 1:S2-8."
quote:
Random drift, while indifferent to the functionality of the molecular features on which it acts, may nevertheless affect evolving molecular mechanisms. It can lead to functional novelty in either gene structure or regulation. In particular, a nearly neutral (in the sense of Ohta), somewhat deleterious mutation can result in a loss of efficiency in gene regulation, and this loss is expected at times to be compensated by a selected event of a particular type: the use of an additional regulatory factor. An accumulation of additional regulatory factors, implying a combination of events of drift and selection, can permit regulatory systems to achieve an increase in both specificity and complexity as mere byproducts of a particular repair process. Nearly neutral mutations thus may, at times, constitute a required pathway for increases in gene interaction complexity. The process seems to point to an inbuilt drive-built into the gene interaction system itself-toward the evolution of higher organisms. This is a matter worthy of experimental exploration, since the general foundations for the evolution of "higher" from "lower" organisms seems so far to have largely eluded analysis.
or
quote:
Abstract. The neutral theory often is presented as a theory of “noise” or silent changes at an isolated “molecular level,” relevant to marking the steady pace of divergence, but not to the origin of biological structure, function, or complexity. Nevertheless, precisely these issues can be addressed in neutral models, such as those
elaborated here with regard to scrambled ciliate genes, gRNA-mediated RNA editing, the transition from selfsplicing to spliceosomal splicing, and the retention of duplicate genes. All of these are instances of a more general scheme of “constructive neutral evolution” that invokes biased variation, epistatic interactions, and excess capacities to account for a complex series of steps giving rise to novel structures or operations. The directional and constructive outcomes of these models are due not to neutral allele fixations per se, but to these other factors. Neutral models of this type may help to clarify the poorly understood role of nonselective factors in evolutionary innovation and directionality.
Evolutionary innovation....
On the Possibility of Constructive Neutral Evolution Arlin Stoltzfus J Mol Evol (1999) 49:169–181
and finally "Neutral Mutations and Punctuated Equilibrium in Evolving Genetic Networks
Stefan Bornholdt1 and Kim Sneppen"
and a thought provoking Santa Fe working paper
quote:
The Evolution of Evolvability in Genetic Linkage Patterns
John W. Pepper
Abstract:
A number of factors have been proposed that may affect the capacity for an evolutionary system to generate adaptation. One that has received little recent attention among biologists is linkage patterns, or the ordering of genes on chromosomes. In this study, a simple model of genetic interactions, implemented in an evolutionary simulation, demonstrates that clustering of epistatically interacting genes increases the rate of adaptation. Moreover, long-term evolution with inversion can reorganize linkage patterns from random gene ordering into this more modular organization, thereby facilitating adaptation. These results are consistent with a large body of biological observations and some mathematical theory. Although linkage patterns are neutral with respect to individual fitness in this model, they are subject to lineage level selection for evolvability. At least two candidate mechanisms may contribute to improved evolvability under epistatic clustering: clustering may reduce interference between selection on different traits, and it may allow the simultaneous optimization of different recombination rates for gene pairs with additive and epistatic fitness effects.
[ 26. August 2003, 00:07: Message edited by: Pim van Meurs ]
IP: Logged
|
|
Jack Foster
Member
Member # 79
|
posted 26. August 2003 00:56
Thanks for the links Pim! I've read the articles since you've provided the links before, but they do belong on this thread.
quote:
I would be interested though to hear what the ID perspective on this issue would be?
I'm not capable of providing THE ID perspective! I'm just an interested spectator. I'll note that designers of evolutionary systems do not focus solely on selecting fit individuals; they design variability into their systems and they provide robustness by allowing for neutral evolution. That's just good design. I'm not sure if evolvability is even possible without neutrality.
We are our own best model. Scientists design evolutionary systems, and they will one day begin to dabble with the bio-chemical platform. Life's evolution is so far beyond what our evolutionary designers are capable of. All of this leads me to believe that life had a starting place that included the universal genetic code, but quite frankly I would be even more impressed with a Designer who created the conditions for the advent of life from the Big Bang. The evolution of evolvability makes this a possibility, but the speculations that I've seen so far seem far-fetched. Perhaps the molecular machines of life are a necessary requirement for evolution. The bottom-line is that we don't know, but as good detectives we should be open-minded to both possibilities. And in either case we should continue to ask "how is this possilbe?": Reverse Engineering!
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 26. August 2003 02:15
Jack: I'm not sure if evolvability is even possible without neutrality.
I am not sure either although I could think of scenarios which do not require neutral evolution such as hypermutation in response to stress. I find it fascinating though to see how neutrality, once somewhat of a 'stepchild' of evolution, seems to gain some strength. In fact, it makes a lot of sense since neutrality can explore genotype space and add variability in genotype space.
The missing link?...
Certainly combined with scale free networks neutrality and selectable evolution seem to provide many of the answers as to evolvability, modularity, degeneracy and robustness. The fact that at least in RNA space, neutral pathways extend all throughout genotype space. A scale free distribution characterized by close proximity in genotype space seems to be helpful in explaining the success of mutation and selection, punctuated equilibrium (exploration of neutral space followed by innovative selectable evolution).
I love this topic
IP: Logged
|
|
Mesk
Member
Member # 630
|
posted 28. August 2003 03:20
Pim,
I guess you've been following the work of Michael Lynch? He gave a talk at the International Congress of Genetics on the evolution of genomic complexity - basically an extension of his duplication-degeneration-complementation theory for the maintenance of duplicate genes to account for other genomic features. He sees genomic complexity as arising firstly through slightly deleterious mutations (e.g. duplications, intron insertions), which are then exploited by natural selection for adaptive purposes. He argues that this process is only likely to occur in organisms with small population sizes, where selection against increased genome size and repeat insertion events is much weaker - explaining why complexity can evolve in large organisms, but is selected against in prokaryotes and unicellular eukaryotes.
Complexity itself arising non-adaptively - it's an interesting idea. It might even be true. ![[Smile]](smile.gif)
Mesk.
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 28. August 2003 12:55
Mesk wrote quote:
He argues that this process is only likely to occur in organisms with small population sizes, where selection against increased genome size and repeat insertion events is much weaker - explaining why complexity can evolve in large organisms, but is selected against in prokaryotes and unicellular eukaryotes.
That seems to suggest (or even imply with necessity? I'm not sure) that once metazoa arise, there'd be a cascade of diversification and complexification. Am I misconstruing it? I've not followed Lynch at all, so I'm groping here. Any refs I might take a look at?
RBH
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 28. August 2003 13:16
Lynch Homepage
quote:
The Evolution of Genome Complexity. Despite the general view that a causal link exists between complexity at the genomic and organismal levels, little thought has gone into the mechanisms that are responsible for the origin of the fundamental features of the eukaryotic genome. Using population-genetic principles as a guide to understanding the evolution of duplicate genes, introns, and regulatory-region complexity, our work is advancing the hypothesis that much of genomic complexity initially evolved as a passive indirect response to reduced population sizes in eukaryotes. One of the primary goals of our work on gene duplication is to explain the shortcomings of the classical model, which postulates that the usual fates of duplicated genes are either the conversion to a nonfunctional pseudogene and or the acquisition of a new function. We believe that duplicate genes are frequently preserved through a partitioning of functions of ancestral genes, rather than by the evolution of new functions. Empirical studies are being pursued to test this hypothesis as well as to estimate the rates of origin and loss of new genes. In addition, we have extended this work to demonstrate that gene duplication may play a central role in the origin of postzygotic reproductive isolation (speciation) by causing passive changes in genetic maps of isolated populations. Our work on intron evolution is focused on the hypothesis that newly arisen introns are typically mildly deleterious and has the goal of understanding how introns eventually came to be integrated into fundamental aspects of gene-transcript processing.
The Evolutionary Fate and Consequences of Duplicate Genes
NUCLEOTIDE SUBSTITUTIONS AND THE EVOLUTION OF DUPLICATE GENES
Supplementary data
New Scientist Article: Generating robustness
quote:
Gene duplication helps to create genetic robustness against mutation.
By Jonathan Weitzman
Gene knockouts often fail to reveal phenotypes, suggesting that biological systems are laden with compensation mechanisms, which might involve functional redundancy between duplicated genes or between alternative pathways and networks. In the January 2 Nature, Zhenglong Gu and colleagues describe a genome-wide evaluation of genetic robustness against null mutation (Nature, 421:63-66, January 2, 2003).
Analysis of fitness measurements for nearly all single-gene deletion mutants in the yeast Saccharomyces cerevisiae, showed that duplicate genes were less often associated with lethal phenotypes, and that deletion of duplicate genes had similar fitness effects. Gu et al. found evidence for a correlation between the sequence similarity of the duplicates and the frequency of compensation.
In addition, deleting the gene copy that is most highly expressed had the greatest effect on fitness. Functional compensation between duplicate genes may account for a quarter of gene deletions without phenotypes.
Speaking of gene duplications, from my understanding of Behe's interview, Behe himself is getting involved in studying gene duplications as well.
quote:
My current work is an attempt to model the evolution of new protein functions through gene duplication. Gene duplication is purported to be a major pathway for the Darwinian evolution of biochemical novelty. However, as in other areas, Darwinists have not closely examined whether gene duplication can realistically do all that they ascribe to it. I hope to help them out in this area by asking those questions.
I find it somewhat fascinating to hear Behe state that "Darwinists have not clesely examined whether gene duplicaiton can realistically do all that they ascribe to it" since I have come across quite a bit of research which seems to be doing just that.
Behe suggests that "To a Darwinist gene duplication followed by natural selection is an easy explanation"
It is exactly those 'elegant' explanations which may help us understand the prevalence of gene duplications in the genome. Behe's objection? "The hitch, as always, is that Darwinists virtually never explain in any detail how natural selection would actually get from protein A to protein B after the gene for protein A duplicated. ". The detailed pathway of course can be quite challenging but as I understand the research some examples have been given in which detail was added based on further scientific inquiry.
For instance
Evolution of antifreeze glycoprotein gene from a trypsinogen gene in Antarctic notothenioid fish
quote:
Freezing avoidance conferred by different types of antifreeze proteins in various polar and subpolar fishes represents a remarkable example of cold adaptation, but how these unique proteins arose is unknown. We have found that the antifreeze glycoproteins (AFGPs) of the predominant Antarctic fish taxon, the notothenioids, evolved from a pancreatic trypsinogen. We have determined the likely evolutionary process by which this occurred through characterization and analyses of notothenioid AFGP and trypsinogen genes. The primordial AFGP gene apparently arose through recruitment of the 5′ and 3′ ends of an ancestral trypsinogen gene, which provided the secretory signal and the 3′ untranslated region, respectively, plus de novo amplification of a 9-nt Thr-Ala-Ala coding element from the trypsinogen progenitor to create a new protein coding region for the repetitive tripeptide backbone of the antifreeze protein. The small sequence divergence (4–7%) between notothenioid AFGP and trypsinogen genes indicates that the transformation of the proteinase gene into the novel ice-binding protein gene occurred quite recently, about 5–14 million years ago (mya), which is highly consistent with the estimated times of the freezing of the Antarctic Ocean at 10–14 mya, and of the main phyletic divergence of the AFGP-bearing notothenioid families at 7–15 mya. The notothenioid trypsinogen to AFGP conversion is the first clear example of how an old protein gene spawned a new gene for an entirely new protein with a new function. It also represents a rare instance in which protein evolution, organismal adaptation, and environmental conditions can be linked directly.
and the comment article
Origin of antifreeze protein genes: A cool tale in molecular evolution
I find Behe's statement 'However, they never look very deeply into the matter.' somewhat at odds with the reality.
Some recent work in gene duplications from a variety of viewpoints
Kondrashov et al , Genome Biol. 2002; 3 (2): "Selection in the evolution of gene duplications"
quote:
Background
Gene duplications have a major role in the evolution of new biological functions. Theoretical studies often assume that a duplication per se is selectively neutral and that, following a duplication, one of the gene copies is freed from purifying (stabilizing) selection, which creates the potential for evolution of a new function.
Results
In search of systematic evidence of accelerated evolution after duplication, we used data from 26 bacterial, six archaeal, and seven eukaryotic genomes to compare the mode and strength of selection acting on recently duplicated genes (paralogs) and on similarly diverged, unduplicated orthologous genes in different species. We find that the ratio of nonsynonymous to synonymous substitutions (Kn/Ks) in most paralogous pairs is <<1 and that paralogs typically evolve at similar rates, without significant asymmetry, indicating that both paralogs produced by a duplication are subject to purifying selection. This selection is, however, substantially weaker than the purifying selection affecting unduplicated orthologs that have diverged to the same extent as the analyzed paralogs. Most of the recently duplicated genes appear to be involved in various forms of environmental response; in particular, many of them encode membrane and secreted proteins.
Conclusions
The results of this analysis indicate that recently duplicated paralogs evolve faster than orthologs with the same level of divergence and similar functions, but apparently do not experience a phase of neutral evolution. We hypothesize that gene duplications that persist in an evolving lineage are beneficial from the time of their origin, due primarily to a protein dosage effect in response to variable environmental conditions; duplications are likely to give rise to new functions at a later phase of their evolution once a higher level of divergence is reached
Lots of good references
Gene duplication and its tie in with scale free networks makes for a very powerful theory of protein evolution imho.
Evolving protein interaction networks through gene duplication
quote:
The topology of the proteome map revealed by recent large-scale hybridization methods has
shown that the distribution of protein-protein interactions is highly heterogeneous, with many
proteins having few links while a few of them are heavily connected. This particular topology is
shared by other cellular networks, such as metabolic pathways, and it has been suggested to be
responsible for the high mutational homeostasis displayed by the genome of some organisms. In
this paper we explore a recent model of proteome evolution that has been shown to reproduce
many of the features displayed by its real counterparts. The model is based on gene duplication
plus re-wiring of the newly created genes. The statistical features displayed by the proteome of
well-known organisms are reproduced, suggesting that the overall topology of the protein maps
naturally emerges from the two leading mechanisms considered by the model.
Selection and gene duplication: a view from the genome Andreas Wagner
quote:
Immediately after a gene duplication event, the duplicate genes have redundant functions. Is natural selection therefore completely relaxed after duplication? Does one gene evolve more rapidly than the other? Several recent genome-wide studies have suggested that duplicate genes are always under purifying selection and do not always evolve at the same rate.
[ 28. August 2003, 14:43: Message edited by: Pim van Meurs ]
IP: Logged
|
|
Jack Foster
Member
Member # 79
|
posted 28. August 2003 19:48
Good reference, Mesk on Michael Lynch; and more thanks to Pim for the links.
I want to think some more about the non-random nature of mutations with respect to fitness: Stoltzfus' mutational and systemic biases. I've tentatively conceded that mutations are stochastic and not directed, but I'm interested in the directionality that different mutations provide for evolution. It's apparent that many evolutionists feel that duplication (and divergence) is one of the more productive forms of mutation in providing the evolution of complexity. If this is true, then I would presume that duplication has a stronger bias than other mutations.
Am I right to correlate productivity of mutation with bias? How might we rank mutations in terms of productivity and therefore bias (if I am right)?
quote:
RBH: That seems to suggest (or even imply with necessity? I'm not sure) that once metazoa arise, there'd be a cascade of diversification and complexification. Am I misconstruing it? I've not followed Lynch at all, so I'm groping here. Any refs I might take a look at?
I presume you're talking about the Cambrian here. This line of reasoning begs the question: what caused the metazoan advent?
IP: Logged
|
|
Mesk
Member
Member # 630
|
posted 28. August 2003 22:15
quote:
RBH:
That seems to suggest (or even imply with necessity? I'm not sure) that once metazoa arise, there'd be a cascade of diversification and complexification. Am I misconstruing it?
No, that's pretty much my take on it too. Multicellular organisms have larger nutritional requirements that unicellular critters, so they must have smaller population sizes. Smaller populations decrease the effectiveness of selection against mildly deleterious events like intron insertions and genetic duplications, so these are more likely to become fixed in the population despite weak negative selection against them. The new genomic material then provides fuel for adaptive evolution - for instance, introns have evolved to perform regulatory functions (i.e. as the sites of enhancers and chromatin modification signals, and as the sources of regulatory non-coding RNA molecules) and, perhaps equally importantly, are the basis for the nonsense-mediated decay mechanism which suppresses premature stop codon mutations.
Lynch is most well-known for his duplication-degeneration-complementation (DDC) model for the maintenance of duplicate genes, whereby each member of a recently duplicated gene pair acquires mutations which knocks out specific function, with different functions being affected for each gene. The end result is subfunctionalisation - that is, all of the functions performed by the original gene are still performed, but now two genes are required for this rather than one. This mechanism is only likely to act in organisms with small population sizes (due to population genetic effects explained by Lynch in the DDC article linked below), and it has interesting consequences: duplicates are maintained far more often than would be expected under the old non-functionalisation/neo-functionalisation model for the fate of duplicate genes (a prediction which appears to have strongly supported by analysis of the genome sequences of various metazoans); genome complexity (i.e. gene number) increases passively; and there is increased fuel for the process of gene neo-functionalisation. There are now a few quite well-characterised examples of the DDC mechanism, including a brilliant study by Vicky Prince's team at the University of Chicago on the expression patterns of Hoxa1 and Hoxb1 genes in zebrafish and mouse (which I don't think has been published in its entirety yet).
The intron evolution stuff that Lynch describes is also particularly interesting to me, because I see an apparent convergence between his theories and those of a biologist working at the University of Queensland, John Mattick. Mattick has a major interest in non-coding RNA, which he sees as the major determinant of organismal complexity - I can't really do his ideas justice in a single post, but his 2001 review in Molecular Biology and Evolution is well worth a read. I've seen the guy give two seminars, and I'm essentially a complete convert to his theories. The tie-in between Mattick's and Lynch's work is that Mattick sees intronic RNA as the major source of regulatory signalling information produced by the genome; the production of intronic RNA signals in parallel with protein-coding sequences during gene expression, he argues, allows a form of networking called "controlled multitasking" which allows a massive increase in organismal complexity without a concomitant increase in protein-coding genes. It turns out that the amount of intronic (and other non-coding) RNA produced by a genome scales closely with developmental complexity, whereas (as we all know) the number of protein-coding genes scales very poorly. It may be that the emergence of introns, which allowed parallel ouputs from gene expression, was another fundamental requirement for the increase in developmental complexity seen in the metazoan lineage.
The two most relevant articles from Lynch that I don't think Pim cited are Preservation of duplicate genes by complementary, degenerative mutations and Intron evolution as a population-genetic process (links are to free full-text articles). Both papers provide theoretical models (with various lines of supporting evidence) for non-adaptive evolutionary mechanisms by which genomic complexity can increase.
quote:
Jack Foster:
I presume you're talking about the Cambrian here. This line of reasoning begs the question: what caused the metazoan advent?
If Lynch is correct, the origin of true multicellularity might have been sufficient to drive an increase in genomic complexity, through the population genetic processes described above. And if Mattick is correct, this increase in genomic complexity (and particularly the spread of intron-like sequences) may have been the event which was required for a rapid increase in organismic complexity, because it allowed the same basic building blocks (protein-coding genes) to be employed and regulated in a vast variety of different ways. Thus an increase in organismal size --> a decrease in population size --> a decrease in the strength of selection --> an increase in genomic complexity --> an increase in organismal complexity.
Mesk.
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 28. August 2003 22:31
Jack Foster wrote quote:
I presume you're talking about the Cambrian here. This line of reasoning begs the question: what caused the metazoan advent?
Well, no, I wasn't specifically thinking about the Cambrian, but earlier, before all those hard parts began to be fossilized. As far as I know, metazooans were around a fairish time before the Cambrian. And while it leaves the question of the advent of multicellular organisms unanswered, it wasn't intended to answer that question but to conjecture about what happened thereafter.
In the OP Jack F. wrote quote:
If we think of computer programs as genotypes, we can think of the operation of the programs as phenotypical. You can then imagine the operation of a program as occupying a unique position within phenotype space, and imagine that each phenotype has its own (though perhaps shared) fitness phase space. Since functional programs as traditionally written are intolerant to change, they can be thought of as occupying "fitness space islands" since there's no way to acquire positive adaptations by randomly altering the program. Variations will typically randomize or "break" the program, therefore fitness phase space is without dimension, or at least is dimension-poor.
Interestingly, that's precisely how Lenski, et al.'s Avida simulation defined things: The programs - sequences of assembly language instructions - are thought of as the genotype and the operation of the programs as the phenotype. I'm glad Jack included that "traditionally written" phrase because I don't think the notion that evolved programs generally easily "break" is the case - while the various programs evolved in the Lenski, et al, study lost some functions in their knockout tests, they also occasionally gained functions when some instructions were knocked out. So the observation that traditionally-written programs are brittle, occupying narrow fitness space islands, does not necessarily generalize to evolved programs.
Jack F. further wrote quote:
The way to attain evolvability with computer programs is by the careful design of the data structure called the genotype to phenotype map. (g-p map) A carefully designed g-p map constrains phenotype space in a way that is friendly to evolution; it provides reach across vast (formerly) unfriendly landscape. Genes may be mapped only to functional phenes, so fitness space islands come together to form something larger! Dimensions are added to fitness space.
Again, I think the issue this addresses is not a general property of computer programs, but is narrowly applicable to programs written 'traditionally' by humans. It is not at all clear how it would generalize to evolved programs, not only those evolved in Avida but those evolved in other evolutionary computation research. The Avida people have some remarks on designing computer languages for evolution. While they don't refer specifically to the gene-phene mapping issue, they discuss some of the properties of languages that differentiate those that are more or less evolvable.
Finally, there is some suggestive work on the evolution of robustness here and here, among other places.
I'm generally skeptical of the utility of analogies from human-written computer programs to either evolved computer programs or to biology. Human-written programs are brittle, not because computer programs are inherently brittle but because humans write brittle programs. The "carefully designed g-map" is necessary because of the deficiencies of human-written programs. While understanding the gene-to-phene mapping is obviously of great importance in biological evolution (since it's the phene that's exposed to selection), I don't think it is important for the reason that Jack F. proffers, to make up for the deficiencies of brittle human-written programs.
RBH
P.S. Added in edit: In fact, it's very likely that the Avida shell, the C-language program that runs the simulation, is considerably more brittle than the digital critters - programs - that evolve in the virtual machine within that shell!
P.P.S. Thanks, Mesk. I was writing in another window whilst you were posting, and missed yours until after I posted mine.
[ 28. August 2003, 22:43: Message edited by: RBH ]
IP: Logged
|
|
Art
Member
Member # 179
|
posted 28. August 2003 23:41
A very fast scan of one Lynch paper Mesk linked to raises a question - in it, as far as I can tell, Lynch seems to ignore that fact that, outside of yeast and probably many fungi, exons are the units that are recognized and assembled in the course of splicing (this model has been termed "exon definition", and is exquisitely supported experimentally). I see no reference to exon definition, but I think this mechanism has very significant ramifications vis-a-vis the issues Lynch raises.
If the thread doesn't pass me by, perhaps I'll have time to look at the paper(s) in more detail. It may be fun to come to a new synthesis that takes into acount exon and intro definition in evolutionary contexts.
IP: Logged
|
|
Jack Foster
Member
Member # 79
|
posted 29. August 2003 13:09
Hi Mesk:
quote:
Multicellular organisms have larger nutritional requirements that unicellular critters . . .
I've got meetings all day, and my internet access at home is down til at least Wednesday. (ugh!) And there's so much good stuff on this thread upon which to comment; I really appreciate the way it's shaping up. But no time to get all my thoughts in. So just a quick post now.
Multicellular organisms also must invent a new system to fulfill their nutritional requirements. (They must invent other new systems as well.) The "collective" is hungry, and they all shout "feed me"! Darwinian evolution cannot invent things; it can only act once there is some basic capacity. To me, this is why macro-evolution is different than micro-evolution; Darwinian evolution easily explains micro-evolution.
So certain macroevolutionary events must be chanced into. But as Stoltzfus has pointed out . . . the variability in Biology is not truly random; there is some directionality provided. Are there biases that increase the odds for invention, for specific inventions?
[ 29. August 2003, 13:11: Message edited by: Jack Foster ]
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 29. August 2003 13:55
One more remark on the computer program analogy: To repeat part of the quotation above, Jack F. wrote: quote:
Since functional programs as traditionally written are intolerant to change, they can be thought of as occupying "fitness space islands" since there's no way to acquire positive adaptations by randomly altering the program. Variations will typically randomize or "break" the program, therefore fitness phase space is without dimension, or at least is dimension-poor.
One characteristic of human programs that does not generalize to biological phenomena is the conscious and deliberate effort on the part of human programmers to write tight code - to create programs that have minimal redundancy and in which every instruction counts. On the other hand, in biological genomic systems there's relatively little (not none, but relatively little) pressure to make every instruction count. The pressure for compactness is much less demanding and so there's scant constraint on 'program' length or 'junk' content.
Along those lines, Ofria, et al did a study of the properties of computer languages that make them more or less evolvable. Very briefly, they found (among other things) that in the digital creatures of Avida, whose genomes are assembly language programs, language variants that enabled loose constraints (in part by varying the available "primitives" - instructions) on genome length evolved more robust critters in the sense that the critters were more tolerant of mutations. And in the Avida case, it appears that the ability to add 'junk' instructions is critical. In my own pilot runs with Avida ver 1.3 and 1.6 I've found that whether 'junk' instructions are selectively deleterious (by consuming replication resources) or selectively neutral has a substantial effect on the course of evolution of Avida populations, with the populations in the latter resisting disruption and more rapidly generating the ability to perform new functions.
So at least in part, the robustness and 'evolvability' of a system depends on the encoding of information. That is not to say that the gene-to-pnene mapping is irrelevant, but the analogy from human programs may give it undue salience.
With reference to Jack F's most recent posting, this is territory in which we need to be real clear and careful about what we and the authors we're reading and citing mean by words like "random" and "directed," lest we descend into pointless hassles based on different implicit understandings of what we are saying with those words.
RBH
Edited to correct an infelicity of expression.
In still later edit: Regarding my remarks on the difference between human-written and evolved programs: I've reread the Ofria, et al, paper, and in the conditions that allowed genomic expansion the evolved programs were two orders of magnitude more robust with respect to mutations than the human-written ancestor with which evolution started.
[ 29. August 2003, 15:56: Message edited by: RBH ]
IP: Logged
|
|
|
Printer-friendly view of this topic