|
Author
|
Topic: Royal Truman: Avida, a biologically unrealistic model for neo-Darwinian Theory
|
Royal
Member
Member # 1060
|
posted 08. April 2004 18:12
Reply to Pim van Meurs’ post 9 Feb. 2004
[PIM]: “This may be Truman's perspective but the more interesting goal is to show how mutation and selection can in fact lead to increase in information and complexity and in fact can create IC/CSI.”
[Royal]: Relevant for me is not “how” but “if” mutations and selection could have performed the creative tasks alleged by neo-Darwinists. I do not believe this to be the case at all.
[PIM]: ![[Confused]](confused.gif) “Whether or not these examples are biologically relevant is of lesser concern than addressing claims based on elimination which are based on the assertion that natural processes (regularity and chance) cannot create CSI.”
[Royal]: Is this not a logical contradiction? You are willing to accept a biologically irrelevant example to conclude natural processes can generate CSI and increase information? It is easy to devise a biologically meaningless computer program which purports to do this. Whether natural processes could do so must surely be the relevant issue!
[PIM]: “Actually the highest proportion would be neutral.”
[Royal]: The examples and refences in my essay demonstrate that under selective conditions, small, rapidly reproducing organisms loss genetic material not needed for immediate survival purposes. As more and more superfluous code is removed, the proportion of mutations which are damaging are not longer neutral.
Tom Schneider responds: "Check out the work of Schuster, Stadler, Fontana and others on the neutrality of RNA and Protein space. Their findings contradict your denial I would argue."
[Royal]: Thanks for bringing these to my attention, Dr. Schneider. You wouldn't have something in pdf format or scanned, huh?? ![[Frown]](frown.gif) Your three references are going to be nearly impossible to locate here in the Third World!
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 08. April 2004 19:59
I'm still pressed for time. My surgery is scheduled for Wednesday, but I'll briefly address two of Truman's immediate issues, and offer some observations by means of which one can self-assess one's understanding of the avida platform.
Truman asked quote:
Why compensate the lower mutation rates with gratuitious free genes? How likely are these new "genes" to further the evolution of new functions according to the Avida program? Now compare this probability to real biological organisms: reading frames get scrambled, expression of existing genes often get messed up, etc.
I didn't "compensate," at least not for purposes of the illustration. I was doing that run for a different reason but it struck me as at least tangentially relevant to the question raised. Runs with just the lower mutation rates also produce lineages that perform the various logic functions, albeit substantially more slowly in evolutionary time.
Truman remarked quote:
Correct. This is an inevitable consequence of the unrealistic model structure.
Well, no, it's not inevitable. A number of runs don't evolve all 9 logic functions. It's likely, but not inevitable, and the likelihood appears to vary at least roughly with the number of intermediate functions that earn reproductive resources. That's what one would expect if evolution of code capable of performing more complex functions were coopting code that evolved to perform other functions.
Truman suggested quote:
Secondly, increase the minimal amount of genetic material which is indispensible for the organism. The model must permit this to be destroyed also by mutations also.
Done it, still works. See the following
The self-assessment of one's understanding of the avida platform is to explain the following observations in a demo avida run I am at this moment starting on another machine on my network.
Initial conditions: Use the parameter settings from Lenski, et al. Seed a 60x60 grid with just one single Ancestor that is 50 instructions in length, of which 18 instructions are required for replication and the remaining 32 are null instructions.
Start the run: The Ancestor replicates rapidly (it's running on a fast machine this time), with mutations occurring occasionally according to the mutation rates defined in the genesis file. As the STATS screen shows, the critters have genotypes 50 instructions long.
First observation: As replication continues and as the grid begins to fill with critters, there is a rapid decrease in the average length of the critters' genotypes. By update 250 or so, the average length has decreased to 22.7 instructions from 50 instructions, and the length of the dominant (most common) genotype is just 20 instructions. The 50-instruction Ancestor is still represented in the population, but at a low frequency.
Notice that this observation fully disposes of Truman's criticism having to do with the seeming abundance of "junk" instructions in the Ancestor that is allegedly available for evolution to work on. In an avida run starting from an Ancestor with a lot of "junk" instructions, those "junk" instructions are very rapidly pruned away leaving critters equipped with very nearly only the minimum code necessary to replicate. That's one of the reasons to actually work with an experimental platform, be it digital or wet lab. One is less apt to make mistaken assumptions about its actual operation, as in the case of Truman's mistaken assumption that the "junk" instructions in the Ancestral genotype are available for evolution to work on. They aren't. They're rapidly thrown away.
Continuing the run, the grid fills with critters. At the moment when the grid is just filled with 3,600 critters, the average length of the 3,600 critters' genotypes is 21.2 instructions and the length of the dominant (most frequent) genotype is 20 instructions. The average critter is executing just 17.5 instructions, indicating that the replication code in a number of lineages has already evolved to be tighter than the Ancestor's 18 instructions. There are 1,038 different genotypes on the grid, the 10 most common totalling 1,698 critters, the top 3 totalling 1,396 critters, and the most common genotype having 795 instances. The Ancestor is apparently extinct or at such a low frequency that it does not appear in the summary statistics. (If one analyzed the files dumped to disk it may still be in the population at a low frequency. I'm not going to bother with that for present purposes.)
Continue the run until the first instance of a logic function (in this case AND) appears in the population.
Second observation: On the occasion of the very first occurrence of the performance of a logic function in the population (in this case AND), the average length of the genotypes in the population had increased to 21.7 instructions and the length of the dominant genotype (which had 322 instances in the population) was 23 instructions. (Note that the dominant genotype was not the one that performed AND, though it might have been a member of a lineage descended from the dominant. I didn't do that analysis.)
Now, two self-assessment questions:
1. Provide an evolutionary explanation (in the context of avida's operation, of course) for the initial substantial decrease in genotype length in spite of the selective neutrality of the number of instructions in a genotype. Explain why the dominant genotype length decreased in the early stages of evolution, down to less than the Ancestor's replication code length, when the allocation of SIPs was proportional to the number of instructions in a genotype. Why did such short genotypes evolve, to the point that by the time the grid was filled, the average genotype was reduced down to almost only the length of the replication code of the (now extinct) Ancestor?
2. Provide an explanation, in terms of the operation of the avida evolutionary platform, for the (slight, but real) increase in genotype length from that minimum length to the point of emergence of the first lineage capable of performing a logic function.
The two questions above can serve as a self-assessment of one's understanding of the operation of the avida platform and the evolutionary processes operating within it. If one can answer them successfully, then one has a fighting chance to understand what avida can and cannot tell us about biological evolution. If anyone cares to essay answers, please PM them to me rather than posting them. Let's let this ride a while so people have a chance to evaluate their own understanding of the platform.
As the run goes on, of course, more phenomena occur that admit of evolutionary explanations in terms appropriate to the avida platform. For example, that first lineage - actually, the one individual critter - that initially performed AND almost immediately went extinct, and AND did not appear again for some thousands of updates. Later NAND popped up in a single individual and then disappeared, as did OR_N still later. Only after still more updates did a logic function (NAND, in this case) get enough representatives successfully replicating to become established in double digits in the population. By then the average length of genotypes in the population was 29.3 instructions, and there were 1,490 different genotypes in the population, the dominant (of 29 instructions) having just 151 instances. When genotypes performing a second logic function (OR_N) got into double digits the average length was 29.5 instructions long and the dominant genotype was still 29 instructions long. Sometime later, after several thousand more updates, critters capable of performing OR appeared in the population and increased rapidly in frequency. Just as rapidly, the frequency of critters performing NAND decreased, suggesting that OR (a more replicationally more valuable function) had emerged in a lineage that was performing NAND, coopting (and incidentally discarding part of) the code that was used to perform NAND. (That's the sort of suggestion one can test by analyzing the specific lineages dumped to files.) Following those lineages is fascinating.
Now I'll be gone for some time. Later, folks.
RBH
[ 08. April 2004, 20:04: Message edited by: RBH ]
IP: Logged
|
|
Royal
Member
Member # 1060
|
posted 09. April 2004 03:38
Reply to Rex Kerr’s post on 8 Feb. 2004
quote:
I think this faulty intuition is behind a large part of the claims made by ID proponents--even the biochemists who should know better! It is understandable, though: if you look at a system that is tightly specified through millions of years of refinement, it is sometimes easy to forget how cross-reactive the starting materials are, in general.
[Royal]: Lets also take into account the excellent comments Rex made in the preceding sentences, e.g. (" But biochemistry is not like this; parts are terribly cross-reactive ") etc.
How is selection supposed to act when identical ensembles of genes over a multi-generation lineage result in different polypeptide environments across supposedly identical members, and within the individual member's lifetime?
The evolutionary model requires a large number of initially useless polypeptide sequences to be generated. They interact with other polypeptide chains or well-evolved proteins in all kinds of mostly undesirable manners. How is natural selection going to preserve and fine-tune this proto-gene in the face of overwhelming 'noise'?
[Rex]: "I don't see that Lenski et al. really addressed the relationship between the size of the jumps needed in their system and the size needed in biological systems."
"They aren't addressing whether the convenient rungs exist in biology--they are demonstrating that if you have rungs, they can be climbed."
[Royal]: Exactly. The Avida computer runs are simply the result of their assumptions. Nothing more, nothing less. One can look at the Avida model structure, consider some parameter settings and predict the new logic functions must evolve.
I am arguing that before anything can be claimed about biological evolution via neo-Darwinian Theory a minimum of realism must be demanded.
[Rex]: quote:
There are very, very good reasons to simplify the problem--for instance, that your simulation will take much too long to run if you don't. Avida was therefore designed to test evolution of complex operations from simple ones and not to be a faithful model of a small-genome living organism.
[Royal]: Simplifying is ok if one can then extrapolate faithfully to the real world. As you correctly pointed out earlier, virtually none of the functions which are biologically coded for in genetic instructions (the actual organism replication; error correction; decoding apparatus) are present in the Avida model.
Therefor they are not subject to destruction or degradation via mutations. The result is a de facto abnormally high proportion of mutations able to do 'good' things.
Applying the kinds of mutational rates used in the paper (or 1/10th as fast as RBH does) with ever more realistic amounts of necessary coding material introduces a hugh amount of 'noise': lots of evolved lineages in Avida would get lost.
At best, huge amounts of time become necessary for even banal evolutionary improvements (neglecting the other factors I have brought up) and thus cannot be the explanation for what we observe in the biological world.
[Rex]: And if Truman doesn't think Avida has anything to do with abiogenesis, I wonder why he thinks that genomes should be small. Selection for smallness, perhaps?
[Royal]: This is the natural result of genome truncation, where everything not immediately necessary which gets eliminated favors that lineage. This is a phenomenon accentuated in small genomes which reproduce rapidly, which otherwise have the best chances of demonstrating neo-Darwinian Theory notions.
If the natural trend is towards elimination of superfluous genetic material, then Avida's claim of increased complexity and new logic functions, which demands ever larger genomes, is falsified.
[Rex]: Truman hasn't provided evidence that the extreme selection he envisions for highly streamlined genomes is biologically realistic
[Royal]: Sure I did. Extensively. Check the original essay.
[Rex]: Indeed, his assumptions would predict that Mycoplasma is, at this very moment, dying out due to an already minimalized genome and unavoidable accrual of mutations.
[Royal]: Mycoplasma! ![[Smile]](smile.gif) Is this not the proof you were looking for? Superfluous DNA got eliminated (since those functions get provided by the host) from a small, rapidly reproducing organism. As predicted.
It won't die out soon because mycoplasma has the necessary genes still available from a vastly more complex host. The later has all kinds of error repair tools available and has characteristics irrelevant to the Avida conditions.
[Rex]: The interesting criticism (if supportable) would be: their assumptions are so generous--and the development of complex functions requires such generosity in assumptions--that biological systems cannot generate complex functions in anything resembling an Avida-like way.
[Royal]: Can I quote this? Precisely my point. Let me add, that the generosity is not limited to biological systems (e.g., the minimal comlexity of a 64 codon genetic system). I argue that any artifical, fully self-contained replicator (also virus, parasites, etc.) controlled fully by encoded instructions would also be incompatible with the Avida model assumptions.
IP: Logged
|
|
Royal
Member
Member # 1060
|
posted 09. April 2004 08:45
Response to RBH’s post on 8 Feb. 2004
[RBH]: First, one can run the same experiment with an Ancestral genotype of just 22 instructions, eliminating the 'junk' instructions Truman thinks are necessary, and get the same general results as those reported in Lenski, et al.. If that's doubted, download the program and try it.
[Royal]: Thanks for the link. I’m hoping to get back with my own analysis. I don’t understand where your “22 instructions” comes from, though. In the paper the runs begin with 15 required instructions and 35 no-operations (nop-c).
Notice how the link contradicts RBH’s claim that Avida is not meant to reflect biological realisties:
quote:
Avida allows us to study questions and perform experiments in evolutionary dynamics and theoretical biology
[RBH]: RBH As was pointed out earlier in this thread, the high mutation rate mostly serves to speed up the evolutionary process.
[Royal]: I addressed this issue already. Avida provides no possibility for extinction due to external factors (lack of nutrients, environmental changes, etc.). Greater number of necessary generations before novel complexity could evolve increases the changes of extinction before this develops.
As pointed out in my essay, as the number of necessary generations increases (the inevitable consequence of incorporating biological realities in the computer model) more opportunities for genome compaction are provided.
Furthermore, introducing even slight biological realism increases the number of generations needed for even banal increases in new functions to the point that neo-Darwinian Theory cannot be the true explanation.
[RBH]: “Throughout the paper, Truman seems to assume that the 3,600 critters in the avida world stay essentially identical to one another through time, that only one lineage is present at a given time.“
[Royal]: Nothing could be farther from the truth. I argued repeatedly in the essay that lineages with shorter genomes will out-reproduce the longer competitors, ceteris paribus.
[RBH]: “Truman's discussion in this connection of the sparseness of protein families in protein space and of the improbability of finding a folded protein (Sauer reference) on pp 7-9 is mostly irrelevant, for three reasons. First, the Lenski, et al., paper at issue did not set out to show how all phenomena evolved.”
[Royal]: Lamentably, it appears RBH has not understood the probabilistic issue behind the folder proteins matter. More below..
[RBH]:
quote:
”Second, the critique is contaminated by Truman's identification of instructions as codons (or perhaps amino acids), which leads to several non sequiturs and irrelevancies. The authors of the program identify avida instructions as more akin to genes, rather than to codons or amino acids, as Truman suggests..”
[Royal]: This is clearly incorrect. From the paper under discussion, p. 143: quote:
Copy errors caused point mutations, in which an existing instruction was replaced by any other (all with equal probability), at a rate of 0.0025 errors per instruction copied.
Let my point out the obvious:
(1) Point mutations produce new codons not genes!
(2) RBH claims ‘instruction’ = ‘gene’. The author’s statement would mean that an existing ‘gene’ was replaced by any other [‘gene’]. Nonsense.
(3) I hope RBH will not compound his misunderstanding and insist 0.25% genes get replaced on average per Avida organism per generation.
Continuing on p. 143: quote:
Single-instruction deletions and insertions also occurred, each with a probability of 0.05 per genome copied.
RBH cannot serious argue that random mutations eliminate and generate a brand new gene every 20 organisms on average by an internal mutation! My identification of codon or amino acid = instruction is simple biology.
[RBH]: “Third (and most interesting), even if instructions are identified with codons, and the operation performed by an instruction is taken as analogous to protein folding/function, Truman's critique is based on the sparseness of the distribution of appropriately folded proteins in protein space.”
[Royal]: This is badly garbled. No one is identifying instructions with protein folding/function. This makes no real-world sense at all. I cannot fathom what RBH is claiming I am saying.
I am arguing, that in going from very simple organic replicators to more complex ones possessing novel, complex functions, additional genes (i.e., proteins) are needed. Families of protein folding patterns are not linked by single mutations, they have very different designs and are separated by large sequence differences. Furthermore, the vast majority of random polypeptides won’t fold in a sensible, stable and reproducible manner, which is just one requirement for a useful coding gene.
Therefor, the proportion of random DNA or amino acid sequences able to produce something new and useful is infinitesimaly smaller than the Avida scenarios use.
RBH]: “But that's irrelevant. What is relevant is the distribution of proteins in codon space. The distribution of proteins in protein space tells us nothing useful about how proteins are distributed in codon space - how they are distributed as a function of changes in codons. But that's the only distribution of interest. When I'm on the road and getting hungry, I'm not at all interested in the distribution of restaurants in size space; I'm interested in their distribution in geographical space. It's the graph induced by the mapping of codon changes into protein variants that is relevant.
[Royal]: What are you trying to tell us? What is the distribution of protein in codon space, or proteins in protein space? By sequence space scientists mean the distribution of useful candidate sequences (amino acids or nucleic acids) among the ensemble of all random possibilities. This is very, very small. Not irrelevent, but a critical factor which needs to be taken quantitatively correctly into account.
For example, all acceptable permutations of amino acids in a chain of length n, vs. all possibilities of that length whereby any of 20 amino acids (ignoring subsequence AA modifications) could be used at each position.
IP: Logged
|
|
michaelgoodrich
Member
Member # 393
|
posted 09. April 2004 13:57
Kerr writes:
Avida isn't meant to be biologically realistic! It's supposed to (from Lenski et al.'s abstract!) "show how complex function can originate by random mutation and natural selection".
----------------------
If Truman is correct in that it does not expose critical parts of the reproduction infrastructure to the effects (most likely deleterious) of mutation then it could not be said even ostensibly to "show how complex function can originate by random mutation and natural selection"
-Mike Goodrich
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 09. April 2004 17:02
Truman seems not to have read my description above of what happens in an avida run employing a long-ish genotype of which only a portion are required for replication. Referring to Rex's remark regarding 'selection for smallness,' Truman writes quote:
This is the natural result of genome truncation, where everything not immediately necessary which gets eliminated favors that lineage. This is a phenomenon accentuated in small genomes which reproduce rapidly, which otherwise have the best chances of demonstrating neo-Darwinian Theory notions.
If the natural trend is towards elimination of superfluous genetic material, then Avida's claim of increased complexity and new logic functions, which demands ever larger genomes, is falsified.
But that is exactly what happens early in an avida run using the parameter settings of Lenski, et al.: the evolving population rapidly prunes away the "junk" instructions of the Ancestral lineage. I repeat from my posting above: quote:
First Observation: As replication continues and as the grid begins to fill with critters, there is a rapid decrease in the average length of the critters' genotypes. By update 250 or so, the average length has decreased to 22.7 instructions from 50 instructions, and the length of the dominant (most common) genotype is just 20 instructions. The 50-instruction Ancestor is still represented in the population, but at a low frequency.
In another few hundred updates, before the capability to perform any logic functions has appeared, the population is wholly composed of short genotypes. One of the self-assessments for understanding the avida system is to explain why that occurs.
When I suggested that one can run the same experiment but start with an Ancestral genotype of just 22 instructions rather than the 50 that Lenski, et al. started with, Truman responded quote:
Thanks for the link. I'm hoping to get back with my own analysis. I don't understand where your "22 instructions" comes from, though. In the paper the runs begin with 15 required instructions and 35 no-operations (nop-c).
The "22 instructions" comes from editing the initial Creature file supplied with the system download and running the system starting from that edited Ancestral critter. My point is that Truman's complaints that the Lenski, et al., study succeeded in evolving complex critters because their Ancestral critter had lots of "junk" instructions for evolution to work with, and that only a small proportion of the code exposed to mutations was replication code, are both mistaken. One can find similar results starting with shorter critters with little "junk" where it is mostly the replication code that is exposed to mutation. And it is that case that in initial conditions similar to those used by Lenski, et al., an initial Ancestral genotype of 50 instructions rapidly sheds the "junk" instructions so that the population evolves to be composed of critters with much shorter genotypes.
So there is some selection pressure for shorter genotypes even given allocation of SIPs as a function of the number of instructions in the genotypes. One of the self-assessment problems I posed is to explain why that occurs in the avida shell. The second self-assessment problem I posed is to explain how it can come to pass that longer critters bearing code capable of performing logic functions can evolve from a population of those short genotypes in the face of the observed selective pressure for shorter genotypes.
Once again, having the program available reduces mistaken assumptions and inferences about what happens in avida runs under various conditions.
In response to my remarks about the distribution of proteins in codon space, Truman asked quote:
[Royal]: What are you trying to tell us? What is the distribution of protein in codon space, or proteins in protein space? By sequence space scientists mean the distribution of useful candidate sequences (amino acids or nucleic acids) among the ensemble of all random possibilities. This is very, very small. Not irrelevent, but a critical factor which needs to be taken quantitatively correctly into account.
The base triplets of tRNA specify amino acids, and the sequence of triplets specify sequences of amino acids (proteins). The question of interest is how proteins are distributed as a function of changes in codons, so the space of interest is the distribution of proteins in a space whose metric is codon changes. The "distribution of proteins in codon space" then is the 'relatedness' map -- the graph -- of proteins as a function of changes in codons. It's not an exotic notion. And it's not the case that the distribution of interest is "the distribution of useful candidate sequences (amino acids or nucleic acids) among the ensemble of all random possibilities" (emphasis added). What is of interest is the distribution of useful candidate sequences in the ensemble of possibilities described by the (non-random) graph. It is becoming increasingly clear that the relevant distributions are not random - they are not well described by random graphs.
And this really is my last posting on this until after surgery!
RBH
[ 09. April 2004, 17:24: Message edited by: RBH ]
IP: Logged
|
|
Royal
Member
Member # 1060
|
posted 09. April 2004 17:53
Repy to RBH’s post from 1 April, 2004
[Royal]: Thanks for the tip about checking the documentation for how energy is distributed. I’ll check this out!
In any event, larger genomes possessing only more genetic junk, i.e., no extra logic functions, need to be penalized in the Avida model to be biologically relevant.
[RBH]: ”the creators of the avida platform did not intend to create a simulation of a specifically biological system, but rather intended to build a system in which evolution occurs independent of purely biological considerations.”
[Royal]: Fair enough. But the authors explicitly claim the software has biological relevance:
quote:
Avida allows us to study questions and perform experiments in evolutionalry dynamics and theoretical biology
(http://dllab.caltech.edu/avida/)
As I pointed out in ealier posts, the obvious application is in cell biology, where new complex functions supposedly could evolve according to the model proposed. For example, the coding instructions undergo “point mutations”, deletions, insertions, etc. (p. 143)
The Nature articel has been widely cited as evidence neo-Darwinian Theory can produce brand new metabolic functions. Now, we all know the authors have not attempted to developed a detailed model of the cell (a group in Japan is attempting that right now!) nor a detailed evolultionary simulation.
But once again, the authors claim Avida is a biologically useful tool. Here is another quote: quote:
“In closing, digital organisms provide opportunities to address important issues in evolutionary biology.”
(p. 143)
Virtually all of us identified the putative evolution of single cell organisms as the obvious topic and background notion in the minds of the authors. (e.g., “Reproduction is asexual and occurs by binary fission.” P. 139). Then permit me to use this tool and determine its relevance for the single most obvious application!
[RBH]:”Similarly, the Lenski, et al., paper takes a question in organic evolution, the evolution of complex systems, and studies it in an alternative evolutionary system, the avida world
[Royal]: Now, going from no logic functions to one and later additional, more demanding functions, has for most users of the Avida software unknown statistical probabilities. To study real biological systems one would like to set such probabilities as parameters, and not be dependent on mysterious computer relationships virtually no biologist understands.
[RBH]: ”Irreducible complexity is at the core of of the Intelligent Design movement's claims about the unevolvability of certain kinds of systems, and is interpreted by ID proponents to be a signature of design by an Intelligent Agency.
[Royal]: This is certainly a key matter. Therefor, one is forced to determine whether the Avida model captures the evolutionary hurdles biological systems face. I maintain this is not the case, and we have no manner to extrapolate to realistic scenarios. I argued in earlier posts we cannot even determine whether the net effect of mutations would be function creation or degradation. Therefor we are dealing with a false analogy.
[RBH]: ”In my next posting on this topic I may pose some problems or questions that assess understanding of the avida platform in order to sort out who understands it well enough to critique it and whose critiques are therefore worth addressing.
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 10. April 2004 16:13
I don't think Avida instructions are either genes or codons. They are discretized instructions.
Also, while I agree that Avida in many ways isn't a faithful model of biological systems, I do not think that is the point. Experiments with Avida argue against Behe's entire line of reasoning with irreducible complexity--namely that circuitous routes are rare. With Avida, a system we can understand and watch closely, circuitous routes to apparently irreducibly complex functions are not rare compared to direct routes. With biological organisms, neither Behe nor anybody else has a really good idea. As such, statements of belief that "they must be rare" ring hollow.
Now to particulars.
Royal asks some questions about evolution in a complex polypeptide environment with annoyingly high cross-reactivity. I could make some guesses, but the bottom line is that I don't really know how it works, and neither does he. I'm not really interested in making arguments for or against evolution from ignorance, so I won't. My point, though, was that "tornado in a junkyard" arguments weren't simply arguments from ignorance but actually from error--from an incorrect view about the cross-reactivity of components.
(Note that I'm not saying I can't think of ways in which it would work--I can!--but it'd just turn into an "is so", "is not" argument. Not too useful.)
Royal states: "I am arguing that before anything can be claimed about biological evolution via neo-Darwinian Theory a minimum of realism must be demanded."
I agree. But I think Avida has met the minimum required to shoot down some of the arguments against evolution--Behe's, in particular. I don't think the paper claims much more than that.
Royal also claims, "virtually none of the functions which are biologically coded for in genetic instructions (the actual organism replication; error correction; decoding apparatus) are present in the Avida model. Therefore they are not subject to destruction or degradation via mutations. The result is a de facto abnormally high proportion of mutations able to do 'good' things."
But this is clearly wrong as demonstrated by RBH's earlier post. 18 instructions are required for replication, and the most common genotype has 20 or so. Almost the entire genome is devoted solely to replication/essential functions! That is far more than in any free-living organism. (In fact, the starting condition of 15-18 essential instructions out of 50 total is closer to what we observe with knockout studies.)
Royal makes some claims about smallness too, but they also are refuted by the above simulation, so I won't address that section.
quote:
[Rex]: Truman hasn't provided evidence that the extreme selection he envisions for highly streamlined genomes is biologically realistic
[Royal]: Sure I did. Extensively. Check the original essay.
Do you cite studies of replication times for two organisms, one of which has more DNA than another? No.
Do you compare rates of duplications/insertions with deletions and show that the selection against the former and for the latter provide strong pressure towards smaller genomes? No.
Do you estimate the fraction of the lifetime metabolic requirement for an organism that is taken up by DNA replication? No.
You cite experiments where you replicate DNA in vitro, and make theoretical curves assuming that all costs are for replication alone, and without carefully adding rates of insertion/deletion. That's nice, as far as it goes, but it's not sophisticated enough to even be relevant to Avida, much less biological organisms.
Here's a hint. If the minimal genome size is N, and the probability of a deletion is pd and is lethal if you hit any of the N and is fine otherwise; and if the probability of insertion is pi and is lethal a fraction q of the time; and if you get a selective advantage of s per generation for each instruction shorter you are; and if we have M unnecessary instructions, THEN the probability of getting a surviving offspring one smaller is approximately
pd * (M / (N+M)) * (1+s/(pd+pi))
and the probability of getting one one larger is
pi * q * (1-s/(pd+pi))
If we set these two equal, the genome length will be approximately stable at length N+M. This gives
(M/(N+M)) = pi*q*(1-s/(pd+pi)) / (pd*(1+s/(pd+pi))) ~= (pi*q/pd)*(1-2*s/(pd+pi))
Solving for M gives
M = N * (pi*q/pd)*(1-2*s/(pd+pi)) / ( 1 - (pi*q/pd)*(1-2*s/(pd+pi)) )
If we plug in numbers for RBH's AVIDA run, N = 18, pi = pd = 0.05; and the others we don't really know, we get
M = 18 * q * (1-20*s) / ( 1 - q*(1-20*s) )
Let's suppose that s = 1/40, meaning that half the energy is used for replication and half for other stuff; and let's suppose that about 5 instructions are likely to be harmless when inserted (the 3 NOPs and a couple others, depending on context) so q = 0.2. Then
M = 18 * 0.2 * 0.5 / ( 1 - 0.2 * 0.5 ) = 18 * 0.1 / 0.9 = 20*0.1 = 2.
So we expect around two surplus instructions to be hanging around. For an extraordinarily rough calculation, this is remarkably close to what RBH actually observed.
Royal is more than welcome to quote my "interesting criticism (if supportable) would be" passage. The supportable bit is important.
-----------
michaelgoodrich wonders whether Avida's lack of exposure of reproduction infrastructure invalidates it as a model of biological evolution.
In brief: no. Digital organisms in Avida do, actually, handle their own reproduction. That part of the code is very much subject to deleterious mutation.
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 10. April 2004 16:20
Truman: You are willing to accept a biologically irrelevant example to conclude natural processes can generate CSI and increase information?
Of course, there is quite a difference between natural processes (regularities as they are called by Dembski) and biologically relevant examples. It is sufficient to show that the Law of Conservation of Information is erroneous and that does not necessary involve using biologically relevant examples.
In fact ID proponents have yet to show that intelligent designers can in fact generate CSI rather than moving CSI around just like natural processes do. I would argue that in a universe closed to natural causes, intelligent designers are as unable to generate NEW CSI as are chance and regularity processes. All they can do is move it around.
That of course leaves open the question as to what can generate new CSI?
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 10. April 2004 22:06
I would go further and claim that they haven't actually shown that there is any CSI at all.
IP: Logged
|
|
michaelgoodrich
Member
Member # 393
|
posted 12. April 2004 13:13
van Meurs writes:
That of course leaves open the question as to what can generate new CSI?
-----------------------------
A universe which is not closed under natural causes and in which embodied intelligent agents are not merely naturalistic automatons?
[ 12. April 2004, 15:27: Message edited by: michaelgoodrich ]
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 12. April 2004 21:52
Michael: A universe which is not closed under natural causes and in which embodied intelligent agents are not merely naturalistic automatons?
I am trying to understand what you are saying here but it does not really hit home. How do we know that CSI can be created in such a universe? How do we establish that such intelligent agents can create CSI?
These are some relevant questions that I believe have not been addressed satisfactorily.
IP: Logged
|
|
Art
Member
Member # 179
|
posted 12. April 2004 22:56
Rex Kerr noted: quote:
Royal asks some questions about evolution in a complex polypeptide environment with annoyingly high cross-reactivity. I could make some guesses, but the bottom line is that I don't really know how it works, and neither does he. I'm not really interested in making arguments for or against evolution from ignorance, so I won't.
IMO, "annoyingly high cross-reactivity" isn't a very good description of the way proteins are (or evolve). There are many ways to illustrate the possibilities (and specificities) of action in a "world" where low-affinity interactions abound. One interesting one (that, admittedly, has to be worked strenuously) is the first figure of the paper by Bolouri and Davidson ("Transcriptional regulatory cascades in development: initial rates, not steady state, determine network kinetics." Proc Natl Acad Sci U S A. 2003 Aug 5;100(16):9371-6). Work out the derivation for the cooperativity term, and it becomes apparent that remarkable specificity can be attained in a scenario that teems with low-affinity interactions (the sort that antievolutionists, I suspect, hold up as posing problems via cross-reactivity).
This particular issue is a non-starter.
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 13. April 2004 01:44
There's no doubt that you can maintain high-specificity networks with low-affinity components. It's a little less clear to me that we understand how cross-reactivity impacts rates of evolution. Anti-Darwinist arguments usually assume no cross-reactivity, so that evolution is playing golf by dropping balls at random on a golf course--and this argument is clearly flawed.
Still, I would like to see something akin to a graph of predicted rate of evolution vs. interaction specificity before saying whether cross-reactivity was a problem or a blessing for evolution. Unfortunately, Avida isn't a good model for this kind of question.
IP: Logged
|
|
michaelgoodrich
Member
Member # 393
|
posted 14. April 2004 12:02
van Meurs writes:
Michael: A universe which is not closed under natural causes and in which embodied intelligent agents are not merely naturalistic automatons?
I am trying to understand what you are saying here but it does not really hit home. How do we know that CSI can be created in such a universe? How do we establish that such intelligent agents can create CSI?
These are some relevant questions that I believe have not been addressed satisfactorily.
-------------------------------------------------
Well you are certainly entitled to your beliefs; recognize however that new approaches typically suffer from pragmatic difficulties especially in the beginning and for subject matters which are problematic to test (e.g., biological forensics).
Also recognize that it is impossible to "prove" something scientifically though something may be easily disproved although the status qou may be recalcitrant in accepting the result.
My statement was meant to be suggestive, as hopefully you are open to the possibility that I alluded to.
Sometimes we have no choice but to take things as axiomatic or be willing to stipulate them for a time and see where these axioms logically lead us.
cheers,
-Mike Goodrich
[ 14. April 2004, 12:03: Message edited by: michaelgoodrich ]
IP: Logged
|
|
|
Printer-friendly view of this topic