|
Author
|
Topic: Albert D.G. de Roos: Evolution Based on Design-by-Contract
|
Moderator
Administrator
Member # 1
|
posted 03. November 2006 16:47
Evolution Based on Design-by-Contract: Origin of Life through an abiotic double-stranded RNA world
by Albert D.G. de Roos
Abstract: It is generally believed that life first evolved from ssRNA (ssRNA) that both stored genetic information and catalyzed the reactions required for self-replication. By modeling early genome evolution on the engineering paradigm design-by-contract, an alternative scenario is presented in which life started with the appearance of double-stranded RNA (dsRNA) as an informational storage molecule while catalytic single-stranded RNA was derived from this dsRNA template later in evolution. Double-stranded RNA can be formed abiotically by hybridization of oligoribonucleotides that are subsequently non-enzymatically ligated into a double-stranded chain. Thermal cycling driven by the diurnal temperature cycles can replicate this dsRNA when strands of dsRNA separate and later rehybridize with oligonucleotides that are subsequently ligated to reform dsRNA. Temperature-dependent partial replication of specific regions of dsRNA can produce the first template-based generation of catalytic ssRNA, similar to the developmental gene transcription process. Thus, the dsRNA-first scenario can be implemented by gradual processes based on abiotic ligation of oligonucleotides and hybridization of complementary nucleotides. Further transition from a dsRNA to a dsDNA world can be based on minor mutations in template and substrate recognition sites of an existing RNA polymerase. Therefore, by defining evolution as an expanding system of functionalities in which existing interfaces remain intact, the ‘dsRNA first’ hypothesis provides a relatively simple gradual evolutionary scenario for the origin of Life.
To read the entire paper, click here.
IP: Logged
|
|
Poul Willy Eriksen
Member
Member # 1976
|
posted 27. November 2006 12:40
Great paper!
From the abstract:
quote:
Abstract: It is generally believed that life first evolved from ssRNA (ssRNA) that both stored genetic information and catalyzed the reactions required for self-replication. By modeling early genome evolution on the engineering paradigm design-by-contract, an alternative scenario is presented in which life started with the appearance of double-stranded RNA (dsRNA) as an informational storage molecule while catalytic single-stranded RNA was derived from this dsRNA template later in evolution.
Ok, so engineering paradigms can be used for modelling evolution. Does this imply that we can attribute intelligence to non-consciousness?
- pwe
IP: Logged
|
|
Albert de Roos
Member
Member # 1823
|
posted 03. December 2006 16:22
quote:
poul willy: Ok, so engineering paradigms can be used for modelling evolution. Does this imply that we can attribute intelligence to non-consciousness?
Hi Poul Willy, thanks for reacting.
I think we can attribute intelligence to material systems, it's what the people at Artificial Intelligence do. It is a matter how you define intelligence in systems, though. I define it as the ability to react in a goal-oriented manner in unforeseen conditions.
But just as I see 'design' detached from religuous meaning, I also think that 'intelligence' does not necessarily mean that consciousness has been involved.
As far as I know, people in Artificial Life have not come up with truly evolvable machines. From my engineering background, I know it is terribly difficult to create systems that can be continuously extended. We have come up with object-orientation, distributed systems etc that we would certainly consider to be the result of logical and intelligent thinking. If nature would have come up with the same design patterns, what reason would be not to call this system intelligent.
IP: Logged
|
|
Poul Willy Eriksen
Member
Member # 1976
|
posted 04. December 2006 10:42
Hi Albert;
You write:
quote:
I think we can attribute intelligence to material systems, it's what the people at Artificial Intelligence do. It is a matter how you define intelligence in systems, though. I define it as the ability to react in a goal-oriented manner in unforeseen conditions.
Sounds fair enough.
Intelligence then is the ability to select the most favorable behavior in a new situation.
Back when I wrote my Master's Thesis, I was dealing with intelligent search strategies (that is my main reason for being interested in ID), but I found two things:
1) Knowledge is more important than intelligence. That is, you can have an ever so smart inference engine, but without knowledge to feed on, it's not going to run anywhere.
2) An abstract intelligence cannot be defined. Intelligence must be defined with respect to a concrete world.
All in all, experience may substitute for intelligence and even be more important.
Possibly the ability to select the most favorable script among those available in a new situation and the ability to extend the avaible scripts.
In a new situation you don't necessarily have much time to think ![[Smile]](smile.gif)
quote:
But just as I see 'design' detached from religuous meaning, I also think that 'intelligence' does not necessarily mean that consciousness has been involved.
Ok, would that mean that you - at least to some extent - agree with the eliminativists (those who hold that consciousness can be eliminated, even for humans)?
quote:
As far as I know, people in Artificial Life have not come up with truly evolvable machines. From my engineering background, I know it is terribly difficult to create systems that can be continuously extended. We have come up with object-orientation, distributed systems etc that we would certainly consider to be the result of logical and intelligent thinking. If nature would have come up with the same design patterns, what reason would be not to call this system intelligent.
Good point - and a frequent point made by evoluyionists. The apparent design in nature, while a sign of intelligence, is not necessarily a sign of a conscious intelligence.
A counter-argument is, how can goal-seeking or any other kind of teleology be attributed to something that has no consciousness?
A stone doesn't fall to the ground because it has some fondness for the ground.
Intelligence is frequently defined as that which changes probabilities. For instance, it is rather unlikely that various materials from around the world should gather together in the shape of a space rocket on some launching pad, and that said spavce rocket should leave the ground and move into space. Not the most likely event without somebody deciding that it should happen.
Can nature change probabilities?
- pwe
IP: Logged
|
|
Melvin H. Fox
Member
Member # 1684
|
posted 05. December 2006 01:01
Can nature change probabilities?
This is precisely the case made for evolution; that random processes combined with natural selection improve the chances life will continue.
But, as I have argued before, life is the most un-natural of endeavors. The exact difference between a star and a star-fish is intent. A star cares not for its preservation, a star-fish cares for little else. If the star-fish fails to take the intended action of inserting its stomach inside the shell of its dinner, the probability it continues to exist is zero.
The natural course in the universe, due the 2nd law, is death not life.
-Mel
IP: Logged
|
|
Albert de Roos
Member
Member # 1823
|
posted 05. December 2006 04:53
quote:
pwe: Knowledge is more important than intelligence. That is, you can have an ever so smart inference engine, but without knowledge to feed on, it's not going to run anywhere.
So, are you suggesting that 'the system of life', or the self-evolving molecular machine as I like to call it, should have the ability to store knowledge about the historical choices? Would a system that is just prepared for flexibility not enough? I think Dawkins' idea about memes as a means for non-genetic inheritance is quite interesting, but seems limited to higher order organisms. How would knowledge be stored in a general model of evolution?
It is an interesting thought, though. I wonder if past experiences could not be encoded in the DNA, as a sort of back-up. E.g. when environmentally challenged, there could be a couple of body plans or solutions ready in the genome. Couple environmental challenges with reorganization of the genome (or higher mutation rate for instance) and you would have such a system.
quote:
pwe: Ok, would that mean that you - at least to some extent - agree with the eliminativists (those who hold that consciousness can be eliminated, even for humans)?
I would not deny that consciousness can have an enormous effect on evolution, but I view consciousness as part of a higher order system that can be reduced to mechanistic counterparts.
quote:
pve: A counter-argument is, how can goal-seeking or any other kind of teleology be attributed to something that has no consciousness?
It may depend on the nature of the goal that you define. I find mainstream evolutionary theory extremely goal-oriented, but the implicit goal, namely a fitness increase, is in my view dubious (if not plainly wrong) and would require foresight.
It's a difficult question, I am now leaning towards a theory in which there is an intrinsic goal that drives the system to an increase in complexity. Coupled with a 'goal' of keeping exiting functionality intact. Cause or result is the question here.
BTW, I think there is plenty of evidence in favor of a system where probabilities are manipulated by organisms. The 'population' just by increasing the chance of an event occurring may also be a design pattern to influence probabilities.
IP: Logged
|
|
peter borger
Member
Member # 722
|
posted 06. December 2006 09:48
quote:
E.g. when environmentally challenged, there could be a couple of body plans or solutions ready in the genome. Couple environmental challenges with reorganization of the genome (or higher mutation rate for instance) and you would have such a system.
These mechanisms are present in microorganisms as cryptic genetic programs that can be activated by for instance prions in yeast.
And what about the stationary state increased mutations?
The mechanism you propose are existing in biology.
Reorganization of the genome (reshuffling)equivalent with evolution sounds like GUToB, Davison's perscribed evolutionary hypo (PEH), frontloading, or the preformationists ideas.
Plenty of bio-evidence for this view.
IP: Logged
|
|
Poul Willy Eriksen
Member
Member # 1976
|
posted 06. December 2006 10:06
Hi Albert;
You ask:
quote:
So, are you suggesting that 'the system of life', or the self-evolving molecular machine as I like to call it, should have the ability to store knowledge about the historical choices?
As you suggest later, the DNA would be the obvious place to store knowledge. Assuming evolution, DNA is a historical result; evolution happens over time.
quote:
Would a system that is just prepared for flexibility not enough?
But what is flexibility? If you mean an ability to respond advantageously to a new situation, I would still say that it requires an ability to compare that situation to some previous situation and select the best action.
quote:
It is an interesting thought, though. I wonder if past experiences could not be encoded in the DNA, as a sort of back-up. E.g. when environmentally challenged, there could be a couple of body plans or solutions ready in the genome. Couple environmental challenges with reorganization of the genome (or higher mutation rate for instance) and you would have such a system.
Yes, that would be the correct analogy, I would think.
Assume some kind of stress increases mutation rates, then you would at least have a system that can try out new solutions.
The 'junk' DNA may be a kind here, don't you think? A repository of things that worked in the past, and may be pullled out again for a try?
quote:
I would not deny that consciousness can have an enormous effect on evolution, but I view consciousness as part of a higher order system that can be reduced to mechanistic counterparts.
Yes, that's a common assumption, and not easy to refute
But if consciousness is reducible, what is the point of it then? That's where I see a problem - if it's there at all, it must make some kind of difference.
quote:
It may depend on the nature of the goal that you define. I find mainstream evolutionary theory extremely goal-oriented, but the implicit goal, namely a fitness increase, is in my view dubious (if not plainly wrong) and would require foresight.
The implicit goal is a relative fitness increase, more carriers of one allele rather than another. Would that require foresight?
I would say 'no' - it's simply a result.
Dr. Dembski writes in his paper The Conservation of Information: Measuring the Cost of Successful Search. that it costs more to find an optimal search than to do a random search, so without prior infirmation you cannot do anything better than a random search.
But I would suggest that he ignores parallelism here. He has an example with searching for a treasure on an islabd. A random search would be prohibitive, if the island isn't very small. But divide the island into two halves, search the one half yourself and hire some one to search the other half. Recursive application of this idea could lead to a very fast search, but of course you may need to pay more to have the search done than you will get out of the treasure.
But if yoiu don't have to pay anyone, if all just agree on having their own little spot to search, it could be done equally quickly, and those who don't finf the treasure might find something else of equal or even higher value?
Parallelism in evolution would mean that we should look at the entire population, and as such it's good enough, if a sufficient number of organisms make it.
quote:
It's a difficult question, I am now leaning towards a theory in which there is an intrinsic goal that drives the system to an increase in complexity. Coupled with a 'goal' of keeping exiting functionality intact. Cause or result is the question here.
That's exactly the question - is the apparent 'goal' an illusion, a post hoc, propter hoc fallacy?
quote:
BTW, I think there is plenty of evidence in favor of a system where probabilities are manipulated by organisms. The 'population' just by increasing the chance of an event occurring may also be a design pattern to influence probabilities.
Yes, that's my point above with parallelism, if I understand you correctly.
Next step would of course be syncretism
- pwe
IP: Logged
|
|
Albert de Roos
Member
Member # 1823
|
posted 08. December 2006 03:03
quote:
peter borger: These mechanisms are present in microorganisms as cryptic genetic programs that can be activated by for instance prions in yeast.
Hi Peter, I didn't know about these cryptic genes. Sounds interesting, especially since yeast is a eukaryotic cell.
quote:
Reorganization of the genome (reshuffling)equivalent with evolution sounds like GUToB, Davison's perscribed evolutionary hypo (PEH), frontloading, or the preformationists ideas.
I know that there are theories that propose intelligent systems or searches, I was merely pointing to the possibility. I believe, like many on this board, that these kinds of systems are necessary to build an evolving system that is robust and flexible.
Another question is how these systems have evolved.
IP: Logged
|
|
Albert de Roos
Member
Member # 1823
|
posted 08. December 2006 03:32
quote:
pwe: But what is flexibility? If you mean an ability to respond advantageously to a new situation, I would still say that it requires an ability to compare that situation to some previous situation and select the best action.
Natural selction would do the trick. Although I am more inclined to a selection in which the organisms (population-wise) that cannot change and are not flexible will be selected against. So it's more like a pruning of the non-fit. Although it will have been addressed by MET, I wonder why this kind of selection hasn't got more attentention, since it is so powerful. Selecting 1% and go on breeding withg this sample is very efficient.
quote:
The 'junk' DNA may be a kind here, don't you think? A repository of things that worked in the past, and may be pullled out again for a try?
That sounds good. I think that it is also possible that things that didn't work out in the past could be used later. Say for instance, an organism has tried another type of body plan, but failed during evolution and falls back to the original stable body plan. In another run, these partially developed solutions may have reused but in another conformation. So, in a couple of trials, a new body plan may be reached that was impossible to reach in one go.
quote:
Parallelism in evolution would mean that we should look at the entire population, and as such it's good enough, if a sufficient number of organisms make it.
I consider 'evolution by population' a powerful design pattern when building an evolving system and probably existed from the very start of life. In my manuscript, I use populations of dsRNA. How would ypu do it without a population: The dependency on a single organism in a varying environment would never work out. That's why I think the fitness selection of chance genetic mutations in single organisms won't work. You put your entire progress in evolution over millions of years in the hands of a single individual. Either you need efficient ways to spread advantageous mutations to the entire population (selective breeding), or maybe the organisms are such that mutations can happen simultaneously in the entire poulation.
IP: Logged
|
|
Albert de Roos
Member
Member # 1823
|
posted 02. May 2007 06:43
This research has now been published in the peer-reviewed journal Biology Direct. Since the peer review process is open and the reviews are published alongside the article, it gives an interesting insight in the peer review process.
IP: Logged
|
|
Albert Voie
Member
Member # 1941
|
posted 03. May 2007 08:01
Hi Albert,
I like your idea! I am studying service oriented architecture for the time being where the reuse of modules is a hot topic. However, I still struggle to see that unguided processes can produce novel functionality within this framework.
Can you please give me an example of a biological "interface", where it is plausible that small uncoordinated genetic changes can produce novel systems? Or do you picture evolution by other means?
IP: Logged
|
|
|
Printer-friendly view of this topic