Evolutionist: You’re Misrepresenting Natural Selection

_{Cornelius Hunter

December 25, 2011

Intelligent Design}

Share: Facebook; Twitter; LinkedIn; Flipboard; Print; Email

How could the most complex designs in the universe arise all by themselves? How could biology’s myriad wonders be fueled by random events such as mutations? Read more

Comments

Time to update some basic comprehensions: Natural Selection Is Ubiquitous Higgs Particle? Dark Energy/Matter? Epigenetics? These Are YOK! Update Concepts-Comprehension… http://universe-life.com/2011/12/13/21st-century-science-whence-and-whither/ Evolution Is The Quantum Mechanics Of Natural Selection. The quantum mechanics of every process is its evolution. Quantum mechanics are mechanisms, possible or probable or actual mechanisms of natural selection. ================= Universe-Energy-Mass-Life Compilation http://universe-life.com/2012/02/03/universe-energy-mass-life-compilation/ A. The Universe From the Big-Bang it is a rationally commonsensical conjecture that the gravitons, the smallest base primal particles of the universe, must be both mass and energy, i.e. inert mass yet in motion even at the briefest fraction of a second of the pre Big Bang singularity. This is rationally commonsensical since otherwise the Big would not have Banged, the superposition of mass and energy would not have been resolved. The universe originates, derives and evolves from this energy-mass dualism which is possible and probable due to the small size of the gravitons. Since gravitation Is the propensity of energy reconversion to mass and energy is mass in motion, gravity is the force exerted between mass formats. All the matter of the universe is a progeny of the gravitons evolutions, of the natural selection of mass, of some of the mass formats attaining temporary augmented energy constraint in their successive generations, with energy drained from other mass formats, to temporarily postpone, survive, the reversion of their own constitutional mass to the pool of cosmic energy fueling the galactic clusters expansion set in motion by the Big Bang. B. Earth Life Earth Life is just another mass format. A self-replicating mass format. Self-replication is its mode of evolution, natural selection. Its smallest base primal units are the RNAs genes. The genesis of RNAs genes, life’s primal organisms, is rationally commonsensical thus highly probable, the “naturally-selected” RNA nucleotides. Life began/evolved on Earth with the natural selection of inanimate RNA, then of some RNA nucleotides, then arriving at the ultimate mode of natural selection, self-replication. C. Know Thyself. Life Is Simpler Than We Are Told, Including Origin-Nature Of Brain-Consciousness-“Spirituality”*** The origin-reason and the purpose-fate of life are mechanistic, ethically and practically valueless. Life is the cheapest commodity on Earth. As Life is just another mass format, due to the oneness of the universe it is commonsensical that natural selection is ubiquitous for ALL mass formats and that life, self-replication, is its extension. And it is commonsensical, too, that evolutions, broken symmetry scenarios, are ubiquitous in all processes in all disciplines and that these evolutions are the “quantum mechanics” of the processes. Human life is just one of many nature’s routes for the natural survival of RNAs, the base primal Earth organisms. Life’s evolution, self-replication: Genes (organisms) to genomes (organisms) to mono-cellular to multicellular organisms: Individual mono-cells to cooperative mono-cells communities, “cultures”. Mono-cells cultures evolve their communication, neural systems, then further evolving nerved multicellular organisms. Human life is just one of many nature’s routes for the natural survival of RNAs, the base Earth organism. It is up to humans themselves to elect the purpose and format of their life as individuals and as group-members. Dov Henis (comments from 22nd century) ***????? ?????? ?? "?????????", ???? ??????????, ?????????? ?????? ?????????, ?????? ??????? ???? An Embarrassingly Obvious Theory Of Everything http://universe-life.com/2011/12/10/eotoe-embarrassingly-obvious-theory-of-everything/ Tags: brain origin, gravitation, gravitons, lifeevolution, nerved organisms, RNAlifehood, spirituality, universeevolutionDov Henis_{June 10, 2012
June
06
Jun
10
10
2012
11:43 PM
11
11
43
PM
PDT}

Correction: second paragraph, 0.0125 -> 0.025material.infantacy_{January 25, 2012
January
01
Jan
25
25
2012
09:04 PM
9
09
04
PM
PDT}

"Now, the fundamental point: once A changes so much that it becomes unrelated to its original sequence, at that point all unrelated states have more or less the same probability to be reached by a random walk."
I need to spend some time reasoning this through.material.infantacy_{January 25, 2012
January
01
Jan
25
25
2012
05:43 PM
5
05
43
PM
PDT}

"i) So, what we need in our scenario is: i1) a starting sequence i2) a series of modifications that realize a random walk i3) a final result"
I imagine that a homologue, hopeful to mutate into a target sequence, has a much higher probability of drifting away from the sequence than toward it. However, given the right mutation rate and enough trials, we should see some drift closer -- but then the probability goes down for successive attempts. Let's consider a 50% homologue sequence for a 100 amino protein. 50 of the aminos match type and position in the target sequence, so there is a 0.5 probability of a point mutation hitting the right position, and a .05 probability of it hitting the right residue, resulting in a 0.0125 probability of scoring a total win, moving the homologue closer toward the target. However the sample space changes. The new sequence now has 51 matching, and 49 non-matching elements. The next mutation has a slightly higher probability of drifting away. If we're drifting, we're in a remarkably declining probability for finding the function. I'm not trusting my reasoning here, nor my hasty math. But some quick induction leaves me with (50!)*(0.01)^50 *(0.05)^50 = 2.7E-101 chance of success, of the homologue drifting to find the target. If beginning with a 90% homologue (100 residues) it works out to (10!)*(0.01)^10 *(0.05)^10 = 3.5E-27 chance of the homologue drifting to the target. If my numbers turn out to be anything close to proper, then I wonder what they would look like with a 500 residue sequence, or more. This is without considering what advantage positive selection might contribute, if one or more of the transitional sequences happens to fold, function, and confer a selective advantage along the way. We can at least imagine that if some intermediate sequence becomes locked in by NS, that chances would improve, although I couldn't say how much. Perhaps not much, since we still need to account for the transition from A -> F, then from F -> B. We would also need to consider the space of the folding sequences set, and the contextual problem that a protein conferring an advantage in some particular context might not confer any advantage at the specific point in time required. IOW, an intermediate sequence may fold properly, but its function may or may not be useful to the organism at its particular stage of development. This is all just stream-of-consciousness commenting.material.infantacy_{January 25, 2012
January
01
Jan
25
25
2012
05:36 PM
5
05
36
PM
PDT}

"...those 2000 or so protein domain are the essebtial core of biological functio in life as we know it. IOWs, they are very important empirical data that need to be causally explained."
[Just jotting down thoughts as they occur to me. It doesn't happen often. xp] This raises the issue that there are two distinct, putative material mechanisms which need to account for the same ability to design and instantiate functional protein products. The first mechanism, an OOL scenario, sequences and manufactures proteins that must be present in the second, DNA-based replication system. The first system must also give rise to the second in its entirety. So either both, disparately functioning systems produce the same types of functional protein structures, or perhaps there is a third theory, the overarching theory of evolution, which demands that specific, sequenced, folded, functional proteins come about in either system.material.infantacy_{January 25, 2012
January
01
Jan
25
25
2012
03:47 PM
3
03
47
PM
PDT}

Do we have any real-world examples of a mutation coming to fixation in a population of say > 1000? Is there any evidence for a slightly deletrious mutation readily drifting through a population?Joe_{January 25, 2012
January
01
Jan
25
25
2012
03:32 PM
3
03
32
PM
PDT}

Elizabeth: No. I believe you are mistaken about that. As I have said many times, any mutation can readily drift, not certainly only those that will be useful. The point is, each new sequence generated by RV is an attempt. All unrelated states have the same, uniform probability to be reached. All unrelated states have the same, uniform probability to drift. Thjerefore, my computation is perfectly correct.gpuccio_{January 25, 2012
January
01
Jan
25
25
2012
01:42 PM
1
01
42
PM
PDT}

But gpuccio, the reason we Darwinists keep going on about non-simultaneity is that if one mutation precedes another, and drifts through the population, there are vastly more opportunities for the second mutation to occur in an individual that already bears the first! And we see this happening in simulations all the time. Even slightly deleterious mutations can readily drift through a population, and if it is a necessary precursor to some advantageous feature, then the chances of the second mutation occurring in some individual becomes very high. That's why your binomial calculation doesn't work. You need to factor in the probability of the first mutation being propagated through the population, and the size of that population in order to find out how many opportunities there are likely to be for the second one to occur in an organism that already has the first. Really off now! Cheers LizzieElizabeth Liddle_{January 25, 2012
January
01
Jan
25
25
2012
12:50 PM
12
12
50
PM
PDT}

Elizabeth: Just one more clarification about a point I had missed. I think you make a technical error here: And that is what is wrong with your binomial calculations: yes, the probability of two simultaneous rare-ish events is lower than the probability of each separately. But, because of neutral drift, simultaneity is not required, and because of the importance of regulatory genes and their role in maintaining homeostasis (a highly selectable function), a rich pool of neutral variance can become a rich pool of selectable variance with small changes in environmental conditions. Simultaneity is not an issue here. That is a common misunderstanding among darwinists. For the computation of probabilities, it is not necessary that all the variations happen simultaneously. I don't know if that is what you meant here, but I want to specify that, because it is an important point. In my example with dies, in the first case, the importat thing is not that we toss two dies simultaneously. We could better concive the case as a single die with 36 facets. The fact is, if only the final result (B) is selectable, we have to have the final complete result (B, with all its dFSCI) before selection can happen. It is not important (nor likely) that all the variation happens simultaneously: but it must all be present in the end, if the function has to be available for selection. Instead, in the second scenarion, the intermediate can be selected. That makes the probabilistic scenario different, and I believe that my application of the binomial distribution to that situation is correct (but I will accept any correction, if I am wrong). The important point is, that second probabilistic scenario is appropriate only if the first event (the intermediate) is completely selected, both nehatively ("fixed", or conserved) and positively ("expanded", or propagated).gpuccio_{January 25, 2012
January
01
Jan
25
25
2012
11:39 AM
11
11
39
AM
PDT}

Elizabeth: Thank you for your reply. I believe you are the forst who has really read all that I wrote here. My compliments! :) I will try to answer briefly your thre arguments: a) All your observations are correct, but don't change anything. In my modellyng, anything that is not related to the functional sequence that confers the local, biochemical activity, cannot improve its probabilities of emerging through a random walk. That's why I say that, whatever the random effects happening, including the modifications in the genetic and outer environments, none of these effects changes the basic fact that all unrelated sequences have the same probability of emerging thorugh a rnadom walk, and that such probability is certainly lower that 1/n_SP. I invite you to make a simulation for that, if you can (I would not be able). You can tahe the space of decimal numbers with 6 digits (not too big, 10^6). The you fix a starting sequence. Then choose a number of unrelated sequences, sharing no digit, or less than 1 digit, as you like, with the starting sequence. Then try a random walk from the starting sequence, and count how many random single digit variations are necessary to reach each of the unrelated states, like in a Monte Carlo simulation. Do it many time, and compute the mean empirical probability of getting each different unrelated state form the starting sequence. According to my reasoning, that probability will be similar for all unrelated states, and it will be lower than 1/10^6 (because the special subset of related states, sharing digits with the starting sequence, can be reached more easily). Now, you can change all the variables you like, but there is no way any variable will make the sequence, say, 361449, specially probable, unless we choose variables that specifically favour that sequence. There may be vairables that favor that sequence, but they will not be more probable than those that favour any other sequence, so we are again in a random situation and an uniform distribution for unrelated states. b) The asnwer to your second point is easy: what you say is correct, nut here I have really been generous, and conceded a scenario where the gene could change neutrally in any possible way, and any functional state reached would be immediately translated and selected. Please, read carefully my post and you will see that that is what I wrote. Therefore, your arguments to explain how that could be possible do not seem specially relevant. c) Here I disagree. Again, you make a fundamental epistemological error here. You say: "And to make my most important point: this is why we simply cannot infer “design” from low probabilities of an alternative. It is, simply, an argument from ignorance (or even an argument from lack of imagination!)" That's wrong. We are looking for the best explanation. We choose between existing explanation. An explanation that has extremely low probabilities of being true is not a valid explanation at all. You are simply saying that anotgher explanation could be possible. OK. Provide it. Science is made with explici, detailed explanations, not with wishful thinking. Unless and until you can offer an explicit alternative, whose probabilites can be really computed and shown to be acceptable, the only competition is between existing explanations. The only problem is, you are not available to accept design as an explanation, and believe that any so called naturalistic explanation, even is not shown, even if not likely to exust, is better than a design explanation. That is dogma and prejudice, not science. And that dogma and prejudice are the essence of all the "argument from ignorance" statements, so common in the darwinist field.gpuccio_{January 25, 2012
January
01
Jan
25
25
2012
05:49 AM
5
05
49
AM
PDT}

I agree. But this mistake is hardly surprising given their antes here: http://richarddawkins.net/discussions/644554-the-blind-watchmaker-maker/comments?page=7 In that page I argued that Nature could not build a hut and got some flak for it. There are 2 reasons I believe why the abilities of NS is misrepresented: 1) It is implicitly assumed that natural selection will weed out the bad variations and keep the better ones. This is false, since natural selection has no foresight and is under no compulsion to do such. 2) Darwin's understanding of NS (as I recall) was based on Malthus posits in his opus 'An Essay on the Principle of Population' as which has been evidently disproved by the very fact that technology has increased food production. My assumption here is that readers are familiar with Malthus posits and that they have been shown to be of little or no effect.Uyi Iredia_{January 24, 2012
January
01
Jan
24
24
2012
10:48 AM
10
10
48
AM
PDT}

Thanks for going to all this effort, gpuccio! I've read it carefully, and I think I have understood some of your terminology that was confusing me earlier (I assume it is a translation issue): for example, where you used the word "fix" or "fixed" you seem to mean what in English we say "conserved". "Fixed", in population genetics, means something different - that all members of a population have that sequence. Similarly, by "expanded", we'd normally say "propagated", or "increase in prevalence". And where you say "pertinent NS" we'd say "NS", i.e. biased drift, as opposed to unbiased drift (which we'd call drift!)> But that's OK, because I think I see what you are saying, and in particular, I'm grateful for the distinction you are making between "local" function and the effect of the phenotype. While I understand that you think you have been generous to Darwinism, I don't actually think you have (:)) because I think you have made a number of untenable assumptions. But I'd like to focus on just two issues for now: Firstly: Even if I accept for the sake of argument your idea that in general a variant that has a "negative" local effect (the protein does less than it used to, for instance) will tend to have a deleterious effect on the reproductive success of the phenotype, while a variant that has a "positive" local effect (the protein does more than it used to) will be more likely to have a beneficial effect on reproductive success of the phenotype, you seem to have ignored the fact that the effect of a variant on the reproductive success of the reproductive success of the the phenotype is not simply a function of the variant; it is just as importantly a function of the current environment. What is more, what improves reproductive success in one environment may decrease it in the next; and what is still more is that the environment itself is a function of the prevalence of variants in the populations (in the valley of the blind, the one-eyed man is king, but in the valley of the one-eyed kings, the two-eyed man is emperor, and may well eat the kings for lunch). This is why I think it is best to think of NS as a special case of drift (or of "pertinent NS" as a special case of NS, as you put it!), as you do, but you need to add another term; if we think of NS as biased drift, that bias is going to change constantly, not only as a function of external factors but of the evolving (in the sense of change of allele frequence over time) population itself. This means that rather than thinking of a new variant as having an immediate beneficial effect (and risking being lost through drift), it may be better to think of new variants as generally being neutral at the time they appear (and as you say, seriously deleterious ones will be rapidly weeded out by negative NS), a proportion of which will drift to substantial prevalence, where they remain dormant in the population, not doing much, except making the population varied, until the environment changes, and the section of the population with a particular variant starts to do better as a result of that variant, at which point it rapidly becomes much more prevalent. I'd also point out, in the same vein, that the environment also includes the genetic environment, which is also undergoing constant change (through drift), and a sequence that was neutral in one genotype may be highly beneficial in another (gene-gene interaction). For instance, a gene for blue skin might be useless in a furry critter, but highly advantageous in a bald one (example deliberately extreme to make my point :) My second point is more serious and concerns your scenario for inactivated duplicates. We now know that regulatory genes are extremely important, and the degree to which a coding gene is expressed will depend on the state of the regulatory genes (whether switched on or off), which in turn will often be a function of the level of protein in the cell (i.e. homeostasis) - so that two perfectly functional identical genes (original and copy), both making the same protein can both be activated, and yet not produce too much protein. Moreover, it is perfectly possible that an inactivating variant of one of the genes could take the form of its only being rarely "switched on" (i.e. in response to a rare environmental signal). If so, it is now free, as you say, to change (because it isn't doing much), and will be switched "on" occasionally, and make a useless protein, but so rarely that it doesn't get in the way of the phenotype. And so this variant has a sporting chance of being propagated by drift. Not only that, but it will not be highly conserved, and soon there will be lots of variants of it and its protein in the population. Now, change the external (for simplicity's sake) environment so that the switch is operated much more often (perhaps there's more oxygen in the environment, or more sunlight); now, those individuals who have variant proteins that are damaging will die off; those that have variant proteins that are neutral will survive; and any whose variant protein is actually useful, will thrive, and come to dominate the population. Now, I'm not saying that is what occurred, or will occur; I'm merely saying that that, right there, off the top of my head, is a perfectly possibly scenario in which a gene could remain active but not often expressed enough to be important, in one environment, and thus be free to acquire variations in the population which would then be heavily selected by an environmental change that increased expression rate. And that is what is wrong with your binomial calculations: yes, the probability of two simultaneous rare-ish events is lower than the probability of each separately. But, because of neutral drift, simultaneity is not required, and because of the importance of regulatory genes and their role in maintaining homeostasis (a highly selectable function), a rich pool of neutral variance can become a rich pool of selectable variance with small changes in environmental conditions. And we can simulate this very easily, and demonstrate that populations with plenty of neutral variance are much more robust in the face of environmental change than more homogeneous genomes (which is why small populations are in far greater danger of extinction than you might expect, and why "hybrid vigor" is so called). And to make my most important point: this is why we simply cannot infer "design" from low probabilities of an alternative. It is, simply, an argument from ignorance (or even an argument from lack of imagination!) In rich feedback systems (non-linear aka chaotic systems) it is extraordinarily hard to model all the things that might happen, and certainly premature to decide that because routes A, B and C are unlikely that A, B and C are the only non-design routes there are. Not to mention the Texas-sharp-shooter problem of estimating the probability of getting B from C via A1, when we don't know whether Z from A via A2 might have been just as awesome. In other words we simply do not know, and cannot know, the distribution of possible end points under the null of Darwinian evolution, and this is what makes any probability based ID argument invalid. Which is not to say that ID is wrong! Just that it's the wrong way of going about testing the hypothesis, and also means that the grounds for thinking it a reasonable inference right now are invalid, seeing as they are mostly - all? - based on probability - or rather improbability - estimates. At least that's how I see it :) Over to you, and thanks again for your posts :) LizzieElizabeth Liddle_{January 24, 2012
January
01
Jan
24
24
2012
10:06 AM
10
10
06
AM
PDT}

I don’t really think that your comments about Dembski, and Behe, are correct. The point in ID is that the random system imagined by darwinists cannot explain the data, nor even with the introduction of NS. I think Dembski has stated very clearly that he assumes an uniform distribution for the random system, that is the onlt reasonable thing to do. Then, he establishes the UPB (indeed, too generous a threshold) as alimit of what a random system can empirically explain.
Well, I certainly do not agree that assuming a uniform distribution of anything is "the only reasonable thing to do"! Very few things in the world have uniform distributions, phases of wave forms being about the only thing I can think of. So if that is the only "stochastic model" he has proposed, then it is self-evidently inadequate! When I have some probability distribution of results for which I want to know the expected distribution under the null, I very carefully design a stochastic model of the null! And often, even with phase distributions, you find that under the null the distribution is only uniform with infinitely large samples! And in non-linear systems (and Darwinian systems are highly non-linear) you certainly can't make any safe assumptions about what you will see under the Darwinian null, for exactly the same reason that we cannot safely assume (at least in Britain) that if you hang your washing out in the morning, it won't be soaking wet when you get home.
Maybe he has not gone into the biological details (but Behe certainly has).
Please point me to a stochastic model of the null produced by Behe.
Anyway, while waiting to understand better the nature of your objections to Dembski and Behe, I will try to analyze the biological random system for you, and to show that it cannot explain data. I have really already done that, but it could be useful to review the resoning in detail and in order, now that maybe we have clarified some epistemological points.
OK :)Elizabeth Liddle_{January 24, 2012
January
01
Jan
24
24
2012
08:58 AM
8
08
58
AM
PDT}

Elizabeth, Eugene, Joe, or whoever is still in tune: Now, I am perfectly aware that the scenario I have described in the previous post is rather artificial: most of my assumptions will never be exactly that way, and some of them are really unlikely. But the point is: all of them are favorable to the neo darwinian scenario, and many of them simplify the computation. So, please, take this model for what it is: a first attempt at modelling a difficult scenario. I will accept any reasonable correction or proposal gladly. I apologize with my ID friends for havinf probably conceded too much to the adversary, but it is only for the sake of discussion. So, the probabilistic model I will use is the cumulative binomia distribution. I will make first an example with the toss of a die, and then apply the results to our scenario. The probability of getting a six tossing one die once is, as all know, 1:6, about 0.166666667. Let's say we toss two dies, and want to know the probability of getting two sixes. In a single toss, it is obviously 0.166666667^2, that is about 0.027777778. Let's say that getting two sixes is our B. Now, let's say that our probabilistic resources are ten tosses. According to the cumulative binomial dostribution, the probability of getting at least one event of probability 0.027777778 in ten attempts is 0.2455066. Now, our alternative scenario, the intermediate scenario, will be similar to tossing one die ten times, and getting at least two sixes. Again, the binomial distribution tells us that the probability for that event is 0.5154833. So, as you can see, with these numbers the scenario where we get two individual events has roughly doubled the probabilities. Now, I must say that I have made simulations, and the increase of probability depends very much on the number of attempts in relation to the global probability of the first event. However, the probabilities of the first scenario are always lower than those of the second scenario. So, let's apply those concepts to our gene scenario. we need to assume specific numbers to do that. So, as I have proposed in the last post, let's say that the probability of getting B from A thorugh a pure random walk are 2^300 (dFSCI of B of 300 bits). Now, we assume that the probabilistic resources (number of random variations in the whole population in a definite period of time) is of 2^100 (that's about 10^300 states, not a small number at all). So: First scenario: the transition from A to B happens by pure random variation, and has a probability of 1:2^300. The probability of getting B from A at least once, in 2^100 attempts, is, according to the binomial distribution: 6.223017 * 10^-61 Second scenario: the transition form A to A1 is naturally selected, and then the transition from A1 to B happens with the same probability and the same probabilisitc resource, as the effect of selection. The probability of having two events, each of probability 1:2^150, in 2^100 attempts, is, according to the binomial distribution: 3.944307e-31 We have an improvement of 30 orders of magnitude (in base 10). About 99 bits. Again, let's remember however that I have considered almost impossibly favorable conditions. In reality, the gain would be certainly be less than that. So, I have tried to show: a) that the existence of a naturally selectable intermediate can highly improve the probability of a transition. b) That it is definitely possible to cimpute the difference in probability, provided that the intermediate is known and can be evaluated. My main point here is: if neo darwinists want really to try to explain protein domains and other forms of biological information, they must face the reality of probabilistic modelling, and not evade it. The modelling can be done and will be done. And will demonstrate that the neo darwinian model is incapable to explain reality. Well, this is the last post in the series. Any comments?gpuccio_{January 24, 2012
January
01
Jan
24
24
2012
08:28 AM
8
08
28
AM
PDT}

Elizabeth, Eugene, Joe, or whoever is still in tune: So, let’s go on, and possibly finish. We have said that pure negative NS cannot help in the emergence of new protein domains. Indeed, it is an obstacle to that, because it preserves existing function. But positive NS is different. Once a new sequence appaers that is naturally selectable, both negative NS and positive NS can act on it: a) Negative NS will fix the new information, and preserve it. b) Positive NS will expand the new gene to most or all the population. First of all, I want to specify that I am not dealing here with the selection, fixation and expansion of the final target, B. As we find B in the proteome as a protein domain, we giva for granted that, once it emerged, it was selecte. But our problem is how to explain the emergence of B. For that, if we don’t want to rely only on RV, (which would be folly, because B has high dFSCI), ve need an intermediate that is naturally selectable. That does change the probabilistic scenario. Some ID supporters have claimed the opposite, but I believe they are wrong. Now, I want to clarify that for the origin of protein domains no naturally selectable intermediate is known: therefore, the following discussion coul simply be omitted. However, as I believe it is important to know how to model NS, I will do it just the same. It can be applied to any possible selectab le intermediate in a transition. I have already listed the properties that an intermediate must have, if it has to be useful in our scenario: a) It must be a sequence intermediate: IOWs, it must have some homology to A and some homology to B. That, and only that, would make the transition from A to A1 easier than the transition from A to B, and so also for the transition from A1 to B. IOWs. A1 must be in the not-completely-away-from A, and at the same time in the not-completely-away-from B. (I must add here that, on second thought, it is not really necessary that the intermediate share some homology with A: it would be enough that its dFSCI is lower that the dFSCI of B, IOWs that it is easier to reach from A because its function is less complex than the function of B. It must instead share some homology with B). b) It must have a local biochemical function, and that function must have a positive effect on differential reproduction c) It must be translated, and enough integrated in the existing scenario so that its local function can be correctly used. That is important too, because no function is useful if it is expressed too little or too much. Well, in the following discussion I will make some assumptions and concession. 1) I will assume that we are dealing with the scenario of a duplicated, inactivated gene. That is, IMO, the most favorable scenario for neodarwinism, because it bypasses the obstacle of negative NS on A, and allows for free neutral variation to happen. 2) I will grant that, as soon as some naturally selectable sequence is reached by neutral variation, for some miracle the gene is translated again, and fuuly integrated in the existent scenario, so that it may be selected. See how good I am? 3) I will assume that both negative and positive NS can act completely on the new selectable gene: IOW, that once it emerges, it will be, in a short time, both completely fixed and expanded to the whole population. 4) I will grant that the intermediate can go on experiencing neutral RV to reach B, while retaining the function, and therefore the fixation, of the partial functional sequence it already has reached. Granting that, I will not have to go into details about the function of the intermediate (if similar to A, to B, or just different). 5) I will assume that the intermediate is exactly “half way”, at sequence level, between A and B. Therefore, it divides the transition from A to B into two separate transition, each with about half the bits of dFSCI of the whole transition. As you can see. All this assumptions, and concession, are favorable to the neopdarwinian scenario. Some of them are just necessary to make the discussion easier. So, the scenario now is the following: a) We have a duplicated, inactivated A. b) The transition from A to B has a dfSCI of, say, 300 bits. c) We call the intermediate A1. d) A1 is functional and naturally selectable. e) A1 shares part of the sequence with B. Let’s grant that the sequence it shares with B will be fixed, and corresponds to about half of the functional complexity of B. f) The transition between A and B can therefore be attained through two separate transitions: from A to A1, and then from A1 to B. g) The dfSCI of the transition from A to A1 is, say, 150 bits. Therefore, A1 can be reached much more easily, starting from A. h) But what happens when A1 is reached? As said, we assume that it is translated and integrated, and that its function is naturally selectable. We also assume that NS will act optimally on it, and in a short time. Therefore, two things happen: h1) The functional part of the A1 sequence is fixed, and will not change any more. We are also assuming that this part of the sequence is shared with B, to make things as good as possible. h2) A1, that was in the beginning present only in one cell, is expanded to the whole population. Now, let’s stop here for a moment. What has happened? We have got a result, the transition form A to A1, that implies “only” 150 bits of functional complexity. As a result of the necessity mechanisms of negative and positive NS, half of the sequence for B is now fixed, and the following transition, from A1 to B, needs to find “only” the other half of the functional complexity of B. So, let’s say that the second transition has too a dFSCI of 150 bits. This result is ensured by the action of negative NS, that fixes the part we have already attained protecting it from further change. What about the probabilistic resources? Well, in the beginning A1 is represented only in one cell, but because of the action of positive NS it is soon present in the whole population. Therefore, form now on, the probabilistic resources to find B from A1 are the same as they were to find A1 from A at the beginning. As both the probability (the dFSCI of the transition) and the probabilistic resources of the two events are the same, we can model the whole system as follows: A random system where we get twice the same event, with the same probabilty. In the next, and final, post I will show how this situation can be modeled probabilistically, and why it differs from the simple random transition from A to B. IOWs. I will show the probabilistic variation due to the intervention of the necessity mechanism of NS.gpuccio_{January 24, 2012
January
01
Jan
24
24
2012
07:49 AM
7
07
49
AM
PDT}

Elizabeth, Eugene, Joe, or whoever is still in tune: As I don't like to leave things unfinished, I will deal in this last post with the analysis of positive NS. I quote my own definition: a) Positive NS all cases of better differential reproduction connected causally to local function variation (even if it were a negative local function variation). So, the main point in positive NS is that the variation in local biochemical function has a positive causal effect on differential reproduction. I would like to start by saying that IMO the role of positive NS is definitely overemphasized. I believe that RV and negative NS are the principles explaining most of what we can observe (except for design :) ). Positive NS is usually invoked to explain what can only be designed, but in reality it does not explain it at all, because of its intrinsic limitations. So, let's see what is necessary for positive NS to take place: a) A variation in local biochemical function must take place by RV. Positive NS can never take place is no variation in local function is present (IOWs, it does not act if only neutral variation of sequence, without any variation of local biochemical function, takes place. b) The local variation of biochemical, be it positive or negative, must be the cause of a positive variation in differential reproduction. c) That positive variation in differential reproduction must be abkle to express itself as an expansion of the original sequence variation to all, or most, or at least a significant part of the population. We call this phenomenon "expansion" of the local variation. It is important to stress that, while the positive influence of the variation on reproduction is a necessity effect that can be detailed and understood in terms of cause and effect, the final expansion is not always a necessary consequence, and can be "modulated", or even prevented, by random effects (drift, other independent and unrelated variables). For instance, a local variation that has a positive effect on reproduction could be lost early because of drift. In the following discussiion, I will restrict the scenario to the following situations: 1) The local variation is an increase in local function, or the appearance of a new function, and the new, or increased, function has a direct positive effecy on reproduction. 2) The necessity effect is strong ennough to determine, often if not always, a significant expansion of the sequence variation in the population, in spite of random effects. The first assumption should be the most common scenario, and simplifies the discussion. The second assumption means only that I am granting maximum power to positive NS in my discussion, something that darwinists should appreciate :) So, let's go on. Because of RV, an existing local biochemical function increases, or a new local biochemical function appears. That is the starting point. Now, I should remind here that our problem is the origin of new protein domains, and in particular the transition from A to B. B, being a new functional domain, can obviously be expanded by positive NS, but that is not relevant to our discussion. Our discussion is about how B arises. So, the thing that should be selected by positive NS (to help us get B) is sopme sequence that can "bridge" the transition between the unrelated A and B, and change the probabilistic modeling of the transition by introducing a necessity effect. We call that selectable sequence an "intermediate", let's say A1. The simple event of the appearance is different according to which of the original scenarios we take. In the case of a functional gene evolving to another new unrelated functional gene, many difficulties arise. The original function must be lost, sooner or later. So, we must explain how the loss of the original function is compatible with reproductive success. Moreover, the function of A1 could be related to the function of A, or to the function of B, or to neither. In general, all these difficulties make the scenario of direct transition fromm one function to a completely different one scarcely credible, even for darwinists. The duplicated gene scenario is better. Here, one of the two genes can be inactivated sooner or later, and become free to change throrugh not-near-A. Negative NS is no more an obstacle, because the gene is no more functional, and all variation is by definition neutral. But another difficulty arises: if the gene is not functional, it is probably not translated. The reason is simple: is non functional sequences were translated, the cell would be repleted with non functional, non folding proteins, that can scarcely be considered a good way to evolve. Non functional proteins are usually a big problem. So, let's say that the gene is not translated, and it varies, and at some moment it reaches the state A1. OK, now it can be selected. But before, it must be translated again. So, either the cell knows in some way that we have reached a functional sequence in the genome (but how?), or it is simply lucky, and translation is casually reactivated when the functional state is reached. The non coding DNA scenario is similar to the duplicated gene scenario. Anyway, let's ignore these difficulties, and let's say that we have A1. and that A1 is translated. Now, positive NS can finally act. OK, but let's look a moment at A1 before. What males of A1 "an intermediate"? IOWs, what properties must A1 have to be really useful in the transition from A to B? a) It must be a sequence intermediate: IOWs, it must have some homology to A and some homology to B. That, and only that, would make the transition from A to A1 easier than the transition from A to B, and so also fro the transition from A1 to B. IOWs. A1 muist be in the not-completely-away-from A, and at the same time in the not-completely-away-from B. b) It must have a local biochemical function, and that function must have a positive effect on differential reproduction c) It must be translated, and enough integrated in the existing scenario so that its local function can be correctluy used. That is important too, because no fucntion is useful if it is expressed too little or too much. Another important point: if and when a selectable sequence appears, in all cases positive NS does not act alone; it always acts together with negative NS. I will be more clear. A1 appears, amd is expressed. It cofers a reproductive advantage to the cell. At that point: a) Negative NS will tend to eliminate RV that affects the newly acquired function. IOWs it will protect the functional part of A1. We call this effect of negative NS: "fixation". b) Positive NS will tend to expand the original variated clone to most or all the population. That can also be described as negative NS acting on the non variated cells, but the concept is the same, so I will keep my terminology. We call this effect of positive NS: "expansion". So, to sum up, our A1 is fixed and expanded. In the first scenario, that will implly the loss of A. In the other two, that is not the case. Well, I think this post is becoming too long. I will stop here for the moment.gpuccio_{January 18, 2012
January
01
Jan
18
18
2012
06:38 AM
6
06
38
AM
PDT}

It was to Joe, but not exclusively :)Elizabeth Liddle_{January 16, 2012
January
01
Jan
16
16
2012
11:29 AM
11
11
29
AM
PDT}

Elizabeth: I am not sure if that is a comment to what I have written up to now. Anyway, as soon as I can I will go on.gpuccio_{January 16, 2012
January
01
Jan
16
16
2012
11:09 AM
11
11
09
AM
PDT}

Yes, that's why I always describe the evolutionary algorithm as heritable variance in reproductive success. But even that isn't quite accurate, because non-heritable, or temporarily heritable, or culturally heritable phenotypic features may also result in variance in reproductive success and serve to filter the gene pool at the level of the population. For example, it is possible that epigenetic variance may serve to keep the gene pool rich and the population more adaptable to changing environments.Elizabeth Liddle_{January 16, 2012
January
01
Jan
16
16
2012
11:04 AM
11
11
04
AM
PDT}

Joe: That's exactly my point. But I wanted to include a brief discussion about unrelated differential reproduction because Elizabeth, and other interlocutors, usually bring it out in the middle of the discussion, in the form of drift or of other accidental involments of genes in indirect selection. So, I thought it was better to rule those aspects out just from the beginning. I have called the neutral scenario "No NS" to underline that it completes the spectrum of possibilities about NS.gpuccio_{January 16, 2012
January
01
Jan
16
16
2012
10:54 AM
10
10
54
AM
PDT}

Natural selection is defined as differential reproduction DUE TO heritable random mutations. If you have differential reproduction due to something else then you do not have natural selection.Joe_{January 16, 2012
January
01
Jan
16
16
2012
08:39 AM
8
08
39
AM
PDT}

Elizabeth and Eugene: So, we have already said that unrelated differential reproduction does not change the computation of probabilities for some specific output B in a basic random system of biological variation. The probabilities to get an unrelated state will remain the same, whatever unrelated differential reproduction is present in the system (drift, or selection of other genes). But how do the three scenarios of pertinent NS affect the computation of probabilities? That is not always obvious. Let's start with the easiest: No NS does not change the scenario. The probabilities reamin the same. That is rather obvious, becasue no necessity mechanism is acting, and the system remains random. Let's go now to negative NS. It is a very important factor, and it acts all the time. The main effect of negative NS is to remove, at least in part, local function variation that is negative, or positive in the "wrong way". I will stick to the first situation, which should by far be the most common. The effect of negative NS is very obvious when the local variation implies total, or serious, loss of local function, and the local function is really necessary for life or reproduction. In that case, the clone with the variation is removed. In all other cases, it can survive, but it usually reproducts worse, except for rare cases where environmental factors can occasionally expand it (see the case of S hemoglobin and malaria). From the perspective of our random walk towards B, what changes as the result of negative NS? The answer is easy. In most cases, it will be impossible for a functional gene to "walk" out of near-A. Or very difficult. If the gene is no more functional (duplicated and inactivated), or if the starting sequence is non coding, negative NS will have no effect. So, just to be simple, the main effect of negative NS will be to keep the functional information as it is, or not to act at all. Nothing in that improves our probabilities to reach B. I will deal with positive NS in next post, because that is really the crucial point.gpuccio_{January 16, 2012
January
01
Jan
16
16
2012
08:22 AM
8
08
22
AM
PDT}

Elizabeth and Eugene: So, NS is defined as differential reproduction in a population. OK, so we must connect that concept to our original concept of "variation of function" at the local, biochemical level. Let's try to be simple. a) A neutral variation of function can have no causal effect on differential reproduction of its own. It can be in some way connected to differential reproduction, but only indirectly, not because of the variation of function, because indeed no local function has changed. So, for instance, a neutral variation could expand, or be lost, because of drift. But that effect will be random in relation to the function of the sequence itself, because it can affect in the same way all sequences, whatever that function. So, an allele with some neuutral variation could be expanded because it is "linked" to some other selectable functional gene because of its position in the chromosome. But again, this has no relationship with the sequence itself and its function. It is, again, a random effect that has nothing to do with the sequence and its function. What I am saying is that even differential reproduction is a random effect if it has no connection with a causal effect of the genetic variation of the sequence, and with its biochemical function. No effect of that kind will ever favour some specific sequence becasue of its sequence function relationship, which is what we need if we want to change the results of a purely random syste. IOWs, indirect effect due to drift or to the selection of something else will not change the probability of getting our B, because all unrelated states share the same probability, and drift or selection unrelated to some specific sequence does not change that fact. b) A negative local variation will usually favour worse differential reproduction. That will not always be the case, ans I suppose that's where Elizabeth would say that a proces is stochastic. And that's true. But the connection between loss of local function and worse reproduction will usually be a necessity relation, that can be diluted, or changed, or even inverted, by random effects of other variables. A special case would be that of some forms of antibiotic resistance, where what is a partial loss of local function can be a function gain because the functional structure that has negatively (or neutrally) chenged is also the target of the antibiotic (see Behe's considerations on "burniung the bridges". c) Finally, the addition (or optimizaion) of a biochemical function can have a causal positive effect on differential reproduction, or it can have no effect, or it can have a causal negative effect on differential reproduction. That seems intuitive enough, so I will not go into details for the moment. Here too, the cause effect relation can be diluted or changed by random effects. So, to go on, we just give the follwinf definitions: 1) We call unrelated differential reproduction any effect (drift or NS of other genes) that has no cause effect relation to the specific local variation of function. 2) We call pertinent NS all effects on differential reproduction that can be causally connected to the variation in local biochemical function, whatever the connection, and however modified by random effects they can be. In the set of pertinent NS, the only one that interests us in this discussion, we will call: a) Positive NS all cases of better differential reproduction connected causally to local function variation (even if it were a negative local function variation). b) Negative NS all cases of worse differential reproduction connected causally to local function variation (even if it were a positive local function variation). c) No NS (neutral scenarios) all cases where local function variation has no causal relationship with differential reproduction (either because there is no local function variation, or because the variation does not affect reproduction).gpuccio_{January 16, 2012
January
01
Jan
16
16
2012
08:02 AM
8
08
02
AM
PDT}

Excellent, got it.Eugene S_{January 16, 2012
January
01
Jan
16
16
2012
07:25 AM
7
07
25
AM
PDT}

Elizabeth and Eugene: OK, let's go to NS. As I have already said, with great resistance from Elizabeth, NS is a necessity mechanisms that applies to particular situations that can arise in the purely random system of biological variation. I will try to summarize here why it is so. I will refer to existing functional proteins in the proteome. I will call "function" their known biochemical activity, and nothing else. Please, Elizabeth, bear with me about this word use. It will makes things simpler in the discussion. I tale full responsibility of the definition, and will not conflate any other meaning. The important point is that function, as defined, depends strictly on primary sequence. For each known protein function in the proteome, defined as restrictively as possible, we have only a limited set of sequences that allow the function in the final protein. We call that set the functional set for that specific function. Whilt its size is cerianly controversial, there is no doubt that it can in principle be defined. So now we call any variation to an existing functional protein, or any addition of a new functional protein, a "variation of function". This is the first point. Variation at the sequence level in an existing protein, or the addition of a new state in a non functiona sequence, can be the cuase of a variation of function, or not. We call any variation of sequence "neutral" if it is not the cause of any variation of function: it neither modifies an existing function in an existing protein, not adds a new biochemical function that did not exist before. Please, note that at this level I am not considering reproductive success at all. We call any variation of sequence "positive" if it optimizes an existing protein function in the biochemical context where it acts (this would require some precisations, but for the moment let's go on), or, more important, if it adds a new biochemical function. We call any variation of sequence "negative" if causes a reducion or loss of an existing protein function. Let's call these three possibilities "the local biochemical effect". The important point is that this effect strictly depends on the protein sequence, and on the sequence-structure-biochemical function relationship. Therefore, this local effect is completely derived from the laws of biochemistry, and is therefore completely a necessity effect of the sequence variation. Next, we will consider the relationship netween local effect and differential reproduction.gpuccio_{January 16, 2012
January
01
Jan
16
16
2012
07:04 AM
7
07
04
AM
PDT}

Eugene: Yes, 10% homology is more or less what you can randomly expect, but it is referred to the protein sequence and the 20 AAs alphabet. SCOP is a protein database. In SCOP, you have a tool called ASTRAL, which gives you the identifiers of groupings sharing less that some percent homology, or having an E-value >= to some value. The lowest homology threshold accepted is 10%, and that gives you 6311 identifiers. The highest E-value is 10, and that gives you 6094 identifiers. So yes, less than 10% homology, for protein sequences, can be considered a random result, and does not allow any assumption of evolutionary relationship.gpuccio_{January 16, 2012
January
01
Jan
16
16
2012
06:47 AM
6
06
47
AM
PDT}

GP, That is much appreciated. The only other thing (as a clarification for me) I wanted to raise but forgot in my previous post is 10% homology meaning no relation. Homology is just sequence similarity, right? If so, as far as I understand, this is supposedly the min one can ever get due to, so to speak, alphabet limitations. In other words, having just 4 letters in our alphabet, there's a limited number of permutations you can get for words and, consequently, our words are bound to have something in common. Am I right in thinking that it is what you mean?Eugene S_{January 16, 2012
January
01
Jan
16
16
2012
06:24 AM
6
06
24
AM
PDT}

Eugene: Your points are very correct, but they do not change much. Always referring to biological systems, point mutation is probably the most common event, and point mutation will give a state in the near-A set. Movements of whole blocks will still give results in the near-A set, at least in more cases, because large parts of sequence will remain unchanged, and strong homology will still be recognizable. The main event that will allow a sudden "leap" to the not-near-A set would be a frameshift mutation. It is true that the result of a frameshift mutation is somehow determined by the original sequence, but it is still unrelated to the original sequence except for the connection deriving from the shift in codons and the genetic code interpretation of codons. There is no reason to believe that such a kind of relation can have any connection with biochemical functionality. So, a frameshift mutation just will lead to not-near-A in a random point in relation to the function of B. IOWs, we are immediately in the situation where all unrelated results are equiprobable. It is not a case that the only explicit model of "evolution" by a frameshift mutation, the origin of nylonase according to Ono, was completely wrong. I don't think that additions of new bits are problematic. Many mutational events change the length of the sequence. In the computation, I reason for a specified length (usually the final length of B) only because it is easier to model the system. It is certainly possible to extend the reasoning to vairable length sequences, and it would certainly not help the darwinian model, but it is simply too complex for me to do that. Usually, however, the proteins in a family, having the same function, are clustered around some mean length, and can be aligned for that main sequence. That's the length and sequence that is considered, for instance, in Durston's method. A transposon operator is certainly a very interesting agent, especially for the possible intelligent modeling of non coding DNA in order to prepare a new protein coding gene. But if it just moves existing blocks, homology will be largely maintained. My point is: let's start with a clear idea of how to model a purely random biological system, whatever the variation operators acting in it. Then we can model how NS can apply to such a system, and when. My starting point remains: in a biological system, unless NS can operate, all variation is random in respect of the emergence of a new functional state that is completely unrelated, at the sequence level, to already existing functional states. For these unrelated states, the probability of each state to be reached by a purely random walk is approximately the same, and it is necessarily somewhat lower than 1/n_SP, because unrelated states are less probable than related states. This is my starting point. In my next post. I will start modeling the effect of NS on such a system.gpuccio_{January 16, 2012
January
01
Jan
16
16
2012
05:49 AM
5
05
49
AM
PDT}

GPuccio, I believe the computation of probabilities should reflect how you define your neighborhood. A neighbourhood of a state s is a set of states S* reachable from s in one move. A move is clearly a local modification of the configuration of a system in a given state. One can define a neighborhood operator mapping from a given state s to its neighbor states in S*. Now, there can be local search operators which can induce what is called very large neighbourhoods whereby a lot of states are reachable in one move. It all depends how you define your move operators, i.e. what your system is allowed to do in one move. E.g. what you define as not-near for one neighbourhood operator, will in fact be within one move for a different operator. That is in fact the only thing I would highlight at the moment. I am not a biologist so apologies if I get something wrong. The simplest operator I can think of is the point mutator. It is easier for me to think in terms of bit strings, so my point mutator will just invert, delete one bit from or add one bit to the initial string (at least this set of abilities to the best of my knowledge is assumed by Gregory Chaitin in his metalife models). A more powerful operator would move blocks of bits around as whole chunks, which can be thought of as cut & paste. We can also think of copy & paste and bit block inversions. Now if we allow our operator to do all of that in addition to point mutations (and specify a-priori the probability to choose each particular type of move, say, 0.5 for point mutations, 0.4 for black cut & paste & 0.1 for block inversions), we can have a pretty powerful/diverse operator. Correct me if I am wrong but I think that a transposon operator could be an example of the above. It appears that you see additions of new bits are problematic. I am aware of Douglas Axe's work which suggests that once you have some function in such a vast configuration space as is the case with proteins, functionality itself necessarily means 'isolatedness' in the configuration space. I am absolutely fine with that. I am also aware of David Abel's thoughts on this, which I also value. He maintains that genetic information is in fact prescriptive and therefore novel genetic information cannot be generated spontaneously. His argumentation is based on the stark absense of any observations that prescriptive information can ever do so anywhere else in nature. Here I would just say that while I am absolutely happy with Abel's argumentation in principle, the details are a gray area to me. I think that it is quite fuzzy in reality. In any event, please continue your posting.Eugene S_{January 16, 2012
January
01
Jan
16
16
2012
05:09 AM
5
05
09
AM
PDT}

Petrushka: Well, B appears at some point. Either some A becomes B, or B is built from scratch. I am discussing the transition scenario, because it is by far the most accepted, even in darwinian reasoning. The "from scratch" scenario is not better for darwinists, as should be obvious.gpuccio_{January 16, 2012
January
01
Jan
16
16
2012
01:53 AM
1
01
53
AM
PDT}

1 2 3 … 8 Next

You must be logged in to post a comment.

Leave a Reply