From junk to func

[FreezBee]

Radio Netherlands has an interesting article: Of cod and code - New discoveries in junk DNA (http://www.radionetherlands.nl/features/science/060421rf).

Cod fish living in cold water have anti-freeze proteins that encapsulate ice crystals in the fish and thereby prevent freezing from spreading, and

Professor Cheng, who presented her results at last week's Annual Main Meeting of the Society for Experimental Biology in Canterbury, has found that the gene for the cod's antifreeze protein has come from a non-coding region of its DNA known as junk DNA.

Junk DNA is essentially DNA that doesn't code for a protein. Making up the chromosomes are the genes - the bits of the DNA that provide the blueprints for our proteins - and the rest of it, the non-coding or junk DNA. The non-coding DNA is thought of as filler, like the letters in a word search puzzle that won't form a word. And yet it makes up a staggering 98.5 percent of human DNA.

Preserving this rubbish seems an inefficient use of time and resources. Evolutionary pressures should favour creatures with less junk DNA. So its conservation may be because it has functions that we don't yet know. The discovery that genes, or at least an antifreeze gene in cod, can evolve out of junk DNA, should add weight to this idea. But how did Professor Cheng decide that this antifreeze gene in cod comes from scrap-heap DNA?

And further down

The chances of a new gene evolving from junk DNA are so slim because junk DNA is full of stop codons, signals that end a protein. Imagine forming a sentence, with junk DNA providing the letters. If you are lucky, the letters join together in an order that makes words. A stop codon is the equivalent of a full stop, automatically ending the sentence, whether you want it to or not. So is there something special about the cod antifreeze gene that allows it to beat the odds?

"This cod antifreeze gene might be an exception because it consists of a short repetitive sequence that only needs to be duplicated four times to give a fully functioning protein," explains Professor Cheng.

Now, nothing is of course proved here; but the idea is interesting. The junk DNA much like junk from a junk yard may be recycled, and why not? Seen from an evolution/creation pov it's extra interesting, because non-coding DNA is turned into coding DNA. Ask a certain German professor about the meaning of that.


- FreezBee

[Viktor Scott]

Interesting.



But I have never really bought into the whole 'evolution should select against 'junk' DNA' argument. The amount of energy required ro replicate a junk DNA ridden genome is miniscule compared to the amount of energy the typical cell uses every day just to get things across its membrane.

Evolution doesn't 'know' that a smaller genome is 'easier' to repliocate, so why should evolution favor getting rid of 'excess' DNA? Again, never bought that whole argument.

[FreezBee]

But I have never really bought into the whole 'evolution should select against 'junk' DNA' argument. The amount of energy required ro replicate a junk DNA ridden genome is miniscule compared to the amount of energy the typical cell uses every day just to get things across its membrane.

Evolution doesn't 'know' that a smaller genome is 'easier' to repliocate, so why should evolution favor getting rid of 'excess' DNA? Again, never bought that whole argument.

You are quite right here, I would think. Only very fast reproducing organisms, such as bacteria, would benefit significantly from not having to waste time and stuff on reproducing junk DNA, so it's hardly selected against in multicellular organisms, and that means that it is still around as a source for NEW functionality without any simultaneous loss of OLD functionality.


- FreezBee

[NeilUnreal]

The junk DNA much like junk from a junk yard may be recycled, and why not?

It makes sense if you think about the discreet mathematics behind genetic evolution. The fact that a sequence of DNA is not needed or is even harmful to an organism does not negate the fact that the same sequence may be the tested result of many generations of selection. A context in which it is useful again (even if the phenotypic expression is different), might provide enough selective advantage to keep it around.

It is an interesting possibility that some of the mechanisms which we think of as "active" in the genotype, might actually be a side-effect of the desireability of preserving warehouses of sequences while not keeping them from interfering with ongoing processes.

In other words, from the long-term standpoint of the meta-evolution of series of interacting populations, the warehouse may be a primary component. Metaphorically, the active sequences may be more like a small store front which admits and dispenses genetic information from the warehouse, while making enough of an "income" to assure both the survival of storefront and the warehouse.

This makes sense if you look at the mathematics behind very simple bitwise genetic coding systems such as those dealt with by Dr. John H. Holland (http://en.wikipedia.org/wiki/John_Henry_Holland) in his schema theorem (http://en.wikipedia.org/wiki/Holland%27s_Schema_Theorem). But it also makes sense in genetic systems which are "chunky" and use very little coding, like the genetic programming systems of John Koza (http://en.wikipedia.org/wiki/John_Koza).

As a result of a bug, I discovered a similar technique in a genetic algorithm system I wrote a number of years ago. Because of the bug, new "organisms" began life with a fraction of their genetic code being from a random recently deceased organism. When I found and fixed the bug, the rate of adaptation slowed down! Even though the dead organisms supplying the genes had been selected against (perhaps many cycles earlier), they still constituted a useful warehouse of genes for further evolution. Junk DNA may be like acting like this; inactive genes from dead ancestors, carried around as a pre-selected pool of useful fragments.

-Neil

[Roy]

Evolution doesn't 'know' that a smaller genome is 'easier' to repliocate, so why should evolution favor getting rid of 'excess' DNA? Again, never bought that whole argument.

Me neither - the overhead of copying extra DNA is going to be minuscule compared to the effects of almost any non-neutral mutation; the selection pressure to reduce the genome will be nonexistent in mega-celled creatures.

Though having said that, in bacteria it does seem to be a significant pressure; many bacteria have extremely small genomes, with almost no wastage. Some of them even double-read their DNA to get multiple genes from the same sequence.

Roy

[Roy]

The fact that a sequence of DNA is not needed or is even harmful to an organism does not negate the fact that the same sequence may be the tested result of many generations of selection. A context in which it is useful again (even if the phenotypic expression is different), might provide enough selective advantage to keep it around.

Aren't you arguing foresight here? That a sequence might be useful under different conditions doesn't provide any immediate selective advantage.

Roy

[NeilUnreal]

Me neither - the overhead of copying extra DNA is going to be minuscule compared to the effects of almost any non-neutral mutation; the selection pressure to reduce the genome will be nonexistent in mega-celled creatures.

I would think the main selective pressure operating against junk DNA would be the fact that -- should it accidently become active -- it would be more likely to a harmful effect on the phenotype than a beneficial effect (and more so than mere random DNA, since it did once code for something with a phenotypic effect). So, if it's useful to have around, it's also potentially dangerous. There may be very interesting mechanisms which have developed to preserve junk DNA and yet limit its ability to reenter the active genenome.

-Neil

[NeilUnreal]

Aren't you arguing foresight here? That a sequence might be useful under different conditions doesn't provide any immediate selective advantage.

I'm not saying that only sequences which might be useful in the future are preserved, rather, that many sequences which were useful in the past might be preserved. These sequences may never be useful again, however, they are probably more likely to be useful than random sequences created de novo. They may even be as much or more useful than many sequences created by mutation of currently useful sequences, since this would require that both deletion of the old effect and addition of the new effect not damage the phenotype.

So only foresight in the sense that -- though the immediate selective advantage may have disappeared -- future novel conditions are likely to be re-hash of something seen before in the evolutionary history, and the preserved, previously active junk sequences are less likely to have been outright deadly or neutral to the phenotype.

-Neil

[FreezBee]

I'm not saying that only sequences which might be useful in the future are preserved, rather, that many sequences which were useful in the past might be preserved.

Isn't this turning things the wrong way around? Those good ol' sequences that have been gone out of use aren't preserved - rather they haven't been deleted.



These sequences may never be useful again, however, they are probably more likely to be useful than random sequences created de novo. They may even be as much or more useful than many sequences created by mutation of currently useful sequences, since this would require that both deletion of the old effect and addition of the new effect not damage the phenotype.

Yes, that's what we would expect, and may actually be an important point in evolution. The creationists are right that changes to coding sequences are somewhat unlikely to add more than they loose, but switching in non-coding sequences are a no-loss situation. And actually, where Gitt et al. would have a problem. How can information come from non-information?


- FreezBee

[NeilUnreal]

Isn't this turning things the wrong way around? Those good ol' sequences that have been gone out of use aren't preserved - rather they haven't been deleted.

Yes, and that is part of my point: maybe “junk” DNA is in part a meta-evolved mechanism for allowing sequences to go out of service without actually being deleted (or a side-effect of the coding technique itself). Maybe we have been looking at things the wrong way around; we’ve been thinking of the junk DNA as “not worth getting rid of” from a selection pressure standpoint. Perhaps it’s the opposite: junk DNA is worth keeping around, even though not immediately useful.

Junk DNA may be the result of some yet-unrecognized process which prevents selection pressure from totally discarding potentially useful sequences. It's that process or effect we need to look for. Experiments suggest such a process would be useful if it does not unduly increase selection pressure against the phenotypes which carry – but don't use – the junk.


Yes, that's what we would expect, and may actually be an important point in evolution. The creationists are right that changes to coding sequences are somewhat unlikely to add more than they loose, but switching in non-coding sequences are a no-loss situation.

From a research standpoint, I think there are at least three big problems we’ll need to solve to really make progress on the func->junk->func question:

1) Better simulations of RNA/DNA evolution, using better algorithms and more powerful computers.

2) A better understanding of the mathematics and algorithmics behind the processes.

(2 & 3 are important because aggregate measures (e.g. Haldane’s Dilemma) don’t really tell us very much, and may even point us in the wrong direction. Things can look very different in process than in aggregate.)

3) Some kind of truly primitive RNA or DNA. The RNA and DNA of even the simplest organisms (e.g. bacteria) have quite a long history of evolution and existence within a phenotype and ecology. The genotypes, phenotypes, environment, and any meta-evolutionary factors have be co-evolving for billions of years. We don’t really know much about how today's RNA and DNA differ from what would be required for basic self-replication and phenotypic expression.

-Neil

[FreezBee]

Yes, and that is part of my point: maybe “junk” DNA is in part a meta-evolved mechanism for allowing sequences to go out of service without actually being deleted (or a side-effect of the coding technique itself).

:lol: You start to sound really religious here - maybe we should rather say that junk DNA was kept for some reason? Cavalier-Smith found a positive correlation between genome size and cell size, and a larger cell obviously have more stuff to work with. But this may also be what you are hinting at with your "meta-evolved mechanism"?



Maybe we have been looking at things the wrong way around; we’ve been thinking of the junk DNA as “not worth getting rid of” from a selection pressure standpoint. Perhaps it’s the opposite: junk DNA is worth keeping around, even though not immediately useful.

Again, this sounds to me as imposing foresight on selection. Rather, as above, I would think that junk DNA must have been worth keeping around independently of its use as a repository for new genes. But apart from that, I fully agree with you that we start looking at junk DNA more like we look at all that stuff we store away in the attic: some day it'll be just what we need :smile:



Junk DNA may be the result of some yet-unrecognized process which prevents selection pressure from totally discarding potentially useful sequences. It's that process or effect we need to look for. Experiments suggest such a process would be useful if it does not unduly increase selection pressure against the phenotypes which carry – but don't use – the junk.

And since there's hardly any selection pressure against junk DNA in most multicellular organisms, ...



From a research standpoint, I think there are at least three big problems we’ll need to solve to really make progress on the func->junk->func question:

1) Better simulations of RNA/DNA evolution, using better algorithms and more powerful computers.

2) A better understanding of the mathematics and algorithmics behind the processes.

I agree; but if I may say so, I'd prefer point 2 before point 1.



(2 & 3 are important because aggregate measures (e.g. Haldane’s Dilemma) don’t really tell us very much, and may even point us in the wrong direction. Things can look very different in process than in aggregate.)

Again I agree - to me Haldane's Dilemma appears to be more of a puzzle than anything else.



3) Some kind of truly primitive RNA or DNA. The RNA and DNA of even the simplest organisms (e.g. bacteria) have quite a long history of evolution and existence within a phenotype and ecology. The genotypes, phenotypes, environment, and any meta-evolutionary factors have be co-evolving for billions of years. We don’t really know much about how today's RNA and DNA differ from what would be required for basic self-replication and phenotypic expression.

No, we really don't - and that's quite the problem.

A good post, Neil :thumb: I believe to remember that I didn't give you pearls for a post because Roy had done that; but I can't use that excuse this time.


- FreezBee

[NeilUnreal]

You start to sound really religious here…

That’s one of the nice things about being both religious and genuinely interested in mainstream science: you don’t have to be afraid of what science might show you about your religion, and you don’t have to be ashamed of the feelings of religious awe you get from looking at the science. :lol:


Cavalier-Smith found a positive correlation between genome size and cell size, and a larger cell obviously have more stuff to work with. But this may also be what you are hinting at with your "meta-evolved mechanism"?


There’s an unanswered “chicken-and-egg” question lurking in here that I think science needs to get to the bottom of. That is, to what extent might things like junk DNA, characteristics of the DNA coding scheme, etc. represent things which have themselves evolved, vs. to what extent do the innate chemistry and mathematics of those things make them good candidates for evolutionary systems. Some of both, I’m sure, but we need to puzzle out the details.


Again, this sounds to me as imposing foresight on selection.

I’m trying hard not to sound like I’m positing some kind of foresight in the metaphysical sense. I'm not. Not because I’m biased, but because what I mean is – to that the extent a type of foresight is operating – it is based on clearly understandable principles of the mathematics and chemistry involved. The junk represents the results of a primitive, automatic learning process, so any teleology comes from this mechanical learning process.

We know that this process of memorizing genotypic information, yet suppressing it in one or more phenotypic generations is valuable in genetic search of a state space. It would not be surprising if it had been part of the “evolution of evolveability.” On the other hand, we may find that RNA/DNA-based genetics has characteristics which make this type of memorizing simpler, thus making RNA/DNA an ideal base for an evolutionary ecology.

Rather than “imposing foresight on selection,” a better way to phrase what I’m saying might be: “selection imposing foresight on DNA pruning.” It works primarily because the search space is not random with each generation. Many generations after a phenotype has gone extinct, its descendants will face a search space that is somewhat different, to be sure, but far from being random with respect to ancestral search spaces. (This is also one spot where “no free lunch” criticisms of evolution break down.) The foresight comes from the non-randomness of the state space, and so really is a type of probabilistic prediction based on learning, not an ad-hoc metaphysical mystery. (Although in my opinion, the very beauty of the process itself is a metaphysical mystery of great profundity.)

(And thanks for the pearls! :teeth: )

-Neil

[Viktor Scott]

Me neither - the overhead of copying extra DNA is going to be minuscule compared to the effects of almost any non-neutral mutation; the selection pressure to reduce the genome will be nonexistent in mega-celled creatures.

Though having said that, in bacteria it does seem to be a significant pressure; many bacteria have extremely small genomes, with almost no wastage. Some of them even double-read their DNA to get multiple genes from the same sequence.

Roy

Indeed - it seems almost self-evident. Yet many, fairly accomplished folks still bring it up. I was a bit surprised to see Maynard-Smith presenting that argument as recently as 1998.

[Viktor Scott]

I would think the main selective pressure operating against junk DNA would be the fact that -- should it accidently become active -- it would be more likely to a harmful effect on the phenotype than a beneficial effect (and more so than mere random DNA, since it did once code for something with a phenotypic effect). So, if it's useful to have around, it's also potentially dangerous. There may be very interesting mechanisms which have developed to preserve junk DNA and yet limit its ability to reenter the active genenome.

-Neil

Interesting point - but if activating (or reactivating it or whatever) were harmful, it most likely, it seems to me, be lost from the population anyway. Just another source of variation which, like any other, has its pros and cons.

[NeilUnreal]

Interesting point - but if activating (or reactivating it or whatever) were harmful, it most likely, it seems to me, be lost from the population anyway. Just another source of variation which, like any other, has its pros and cons.

Yeah, I'm conflicted on this as well. If -- big if -- junk DNA does prove to be a significant source of re-activated DNA (even that is mere conjecture at this point), perhaps one of several things is happening:

1) There is some uneasy balance between junk which is dangerous to re-activate and that which is not.

2) Only relatively harmless junk is retained; the really bad stuff never getting over the selection hurdles before first reproduction. Or,

3) Some mechanisms, yet unknown, which tip the balance in favor of preserving and re-activating non-dangerous segments.

Again, all this is speculation, and the research in the OP may represent a singular occurrence, not the general case. But it seems worth pursuing further, particulary since artificial genetic systems indicate that it might be useful. Even if it doesn't pan out, it seems a lead worth following if only to close down a dead end.

Good topic, FreezBee; this has been fun to think about.

-Neil

[FreezBee]

That’s one of the nice things about being both religious and genuinely interested in mainstream science: you don’t have to be afraid of what science might show you about your religion, and you don’t have to be ashamed of the feelings of religious awe you get from looking at the science. :lol:

True, science isn't the enemy of religion.



There’s an unanswered “chicken-and-egg” question lurking in here that I think science needs to get to the bottom of. That is, to what extent might things like junk DNA, characteristics of the DNA coding scheme, etc. represent things which have themselves evolved, vs. to what extent do the innate chemistry and mathematics of those things make them good candidates for evolutionary systems. Some of both, I’m sure, but we need to puzzle out the details.

Good point. The evolutionary pathways are not a fixed set of possibilities; but do themselves change over time. We need coding DNA to have junk DNA which is then a repository for new coding DNA; but how do we start the show? Apparently RNA came first - is there any junk RNA?



I’m trying hard not to sound like I’m positing some kind of foresight in the metaphysical sense. I'm not. Not because I’m biased, but because what I mean is – to that the extent a type of foresight is operating – it is based on clearly understandable principles of the mathematics and chemistry involved. The junk represents the results of a primitive, automatic learning process, so any teleology comes from this mechanical learning process.

Ok, I read this to imply that the teleology is part of the process itself, not an outside force pulling evolution. Do I read you correctly?



Rather than “imposing foresight on selection,” a better way to phrase what I’m saying might be: “selection imposing foresight on DNA pruning.” It works primarily because the search space is not random with each generation. Many generations after a phenotype has gone extinct, its descendants will face a search space that is somewhat different, to be sure, but far from being random with respect to ancestral search spaces.

Quite correct. Grasses are supposed to have evolved from plants pollinated by insects, these plants in turn having evolved from plants pollinated by the wind - just as grasses are. Environmental conditions change over time and may change back to something that has existed before.



(This is also one spot where “no free lunch” criticisms of evolution break down.)

Really good point :thumb: Evolutionary algorithms are based on static search spaces, not on search spaces that themselves change - even simply the evolution of one species will change the search space for other species in the same environment.



The foresight comes from the non-randomness of the state space, and so really is a type of probabilistic prediction based on learning, not an ad-hoc metaphysical mystery.

Ok, this explains it better - and you are right: since non-randomness can be considered as information, and information is needed for learning, it all adds up somehow.



(Although in my opinion, the very beauty of the process itself is a metaphysical mystery of great profundity.)

Amen :smile:


- FreezBee