Statistical/probability issues in speciation

[Pyrotex]

...1) Whenever a cataclysm occurs, we eradicate large portions of extant populations, and also sequester small populations

2) Small populations are far more likely to express recessive alleles, simply by arithmetic

3) Ergo, any population that has unexpressed recessive alleles is far more likely to express those phenotypically after a major population reduction or population sequestration, as with a cataclysm.....

Brain flash!! Cosmic insight!!

During the CX, there was another form of cataclysm, catastrophic dispersion. We don't get this today much because all eco-niches are filled. Not so during CX. Evolve an extra thorax section with a pair of legs (a one gene mutation) and you might find yourself King of the Hill by generation 50. Suddenly, the STORM!!! Boom, now you have 100 tiny pockets of survivors, all in new regions scattered all over the archipeligo, with new unfilled eco-niches. Most may breed true and become local Kings of the Hill, but according to 2), you have a magnified chance of RA modifying these sequestered mutants. By generation 100, you might have not just one new proto-species--but a dozen?

[pmaust]

What a neat discussion. I find the whole CX thing very interesting. The climate appears to have been much warmer then and god knows what the chemical makeup of the environment was like. I am waiting for science to reveal its secrets but I suspect that there was something going on that really pushed the evolutionary envelope. ;)

[Biochemist]

Brain flash!! Cosmic insight!!

During the CX, there was another form of cataclysm, catastrophic dispersion. We don't get this today much because all eco-niches are filled. Not so during CX. ...By generation 100, you might have not just one new proto-species--but a dozen?

I think it is hypothetical to suggest that there were more available niches then than now. It could be true, but the number of phyla has decreased since the period just after the CX, by perhaps 50%. Most paleontologists think there were about 70 phyla after the CX. We have maybe 30 now.

 

In any case, having more available niches does not explain how we got viable recessive alleles in the first place. In the generally accepted model, a gene has to be expressed to be selected. If a several genes are required for a new enzyme system (with six being about the average) we still have to figure out how a new enzyme system could be available in a recessive allele to be selected.

 

I still want to talk about the math to make that work. Six or so enzymes in a new enzyme system show up without any apparent previous phenotypical expression. Any suggestions how to get the odds down on this?

[Biochemist]

...Evolve an extra thorax section with a pair of legs (a one gene mutation) and you might find yourself King of the Hill by generation 50.

Incidentally, don't you find it surprising that a single gene mutation can cause such a dramatic change in morphology?

[pmaust]

Incidentally, don't you find it surprising that a single gene mutation can cause such a dramatic change in morphology?

 

I don't know what the rule is on these forums for posting links to other sites but, in 2002 researchers from the University of California Sand Diego uncovered "genetic evidence that explains how large scale alterations to body plans were accomplished during the early evolution of animals". In February 6 of the same year, they published their findings in Nature. I think this discovery would be worth some review here as it might address the question of how drastic changes in morphology can occur from a single mutation.

 

The team discovered that relatively simple mutations in a class of regulatory genes, known as HOX can act as master switches by turning on and off other genes during embryonic development....

 

If someone can aprise me of the rules here, I will be more than happy to post a link to the announcement and summary of this research from UCSD Science & Engineering Press Release.

 

Thanks

 

Paul

[Biochemist]

If someone can aprise me of the rules here, I will be more than happy to post a link to the announcement and summary of this research from UCSD Science & Engineering Press Release.

PM-

 

You can post a link to anything. If the material is copyrighted, you must limit yourself to only a brief excerpt if you want to copy in any of the text.

 

There is a significant body of research on small genomic changes that result in significant changes in morphology. Small (e.g., single codon) genomic changes that result in multiple changes to the body plan are usually referred to as Pleiotropic.

[joel]

To get a functional protein by mutation: you would need at least 200 specific amino acids in specific sequence out of 300 in the protein (it is actually more like 260 on average, but I am making it simpler). This would be randomly 1 in 20^200, or about 1 in 10^260. Heck. To be conservative, let's make it a couple of trillion trillion trillion trillion times more likely, and make it 1 in 10^200.

 

If I read this right, you have calculated a probability for a single mutation 'event' to to move a gene from one protein to another that is 200 amino-acids different. But my text-book has the changes being incremental, usually one codon at a time with an 'evaluation' between each incremental change. Can you clarify what event you're saying has this 1/10^200 likelihood?

 

I wonder if in the earlier days of life, if viruses were not as a rule pathological, and provided a lot of lateral transport for 'chunks' of genes. The dynamics of evolution would be quite different. Just a thought...

[Biochemist]

...But my text-book has the changes being incremental, usually one codon at a time with an 'evaluation' between each incremental change. Can you clarify what event you're saying has this 1/10^200 likelihood?

My hypothetical construct here is that we need to account for evolutionary development of a new enzyme system. I sketched out a "typical" enzyme system that is a half dozen enzymes, each of which requires 200 unique amino acids. This example is a little simplistic, in that mammals probably have dozens to hundreds of enzyme systems that do not exist in invertebrates, but I was starting with a simple case for discussion.

 

An individual mutation could indeed change one codon, but that would not create a functional new enzyme system. Further, a mutated protein could not be selected until is is expressed phenotypically, and have an advantage in its environment.

 

Most enzyme systems to not have functionality until the entire system is present, therefore, there is no function to provide the basis for selection until the entire system is present.

 

Ergo, we get the hypothetical numbers generated above.

[pmaust]

Biochemist, here you go. BTW, this is a great thread!

 

http://ucsdnews.ucsd.edu/newsrel/science/mchox.htm

 

Thanks

[Biochemist]

http://ucsdnews.ucsd.edu/newsrel/science/mchox.htm

This is more of a popular-press point-of-view summary than the source document. But it is still a good addition to the discussion.

 

It is also another example of a presumption of mutation versus a presumption of proscribed behavior. If a single, small edit to the genome results in a significant (and functional) change in body plan, it suggests that the genome contains information load that is not expressed until it encounters a trivial change.

[joel]

Following this slight tangent to the original topic a bit further... There is a ton of Hox literature already. Here is a somewhat more detailed paper that I happen to be aware of: http://www.pnas.org/cgi/content/full/97/9/4504. See figure 4, showing the results of injecting the poor little beatle eggs with some RNA constructed to modulate the Hox gene expression. The larvae look like shrimp rather than beatles, with a dozen extra legs, as also described in pmaust's link.

 

Biochemist comments...

It is also another example of a presumption of mutation versus a presumption of proscribed behavior. If a single, small edit to the genome results in a significant (and functional) change in body plan, it suggests that the genome contains information load that is not expressed until it encounters a trivial change.

 

I don't completely agree with this point of view, except in an abstract context. Each letter in our alphabet has an established meaning. String them to gether to make words with established meanings. Arbitrarily change, add or remove a letter in a word, and you may (a)destroy its meaning; (b)modify its meaning; ©invert its meaning; (d)leave its meaning unchanged, but use a Brittish rather than American spelling convention. Does every word have a latent information load? Do words whose meaning can be inverted with a single letter change have larger latent information loads than words in group (d)? If so, was this planned or designed into language? Just my thoughts on the matter...

 

I'll try to add an on-topic post after I do a bit of (hopefully) intelligent design at work. Cheers!

[Buffy]

Incidentally, don't you find it surprising that a single gene mutation can cause such a dramatic change in morphology?

Going back to the exasting previous thread (in which your opening post here was extracted, and for those of you who're interested, go back n' read it!), it sounds like you still don't buy the notion that genes sure seem to be "modular". From our original discussion, I'm not even sure you understood what I was talking about so I'll take a quick shot at it here:

• Imagine you've got a program for driving a car. Within its code, there's a list of instructions for "turning left" which involves a bunch of stuff--hit signal, look left, turn wheel. There's also a list of instructions for "turning right", involving similarly complex set of steps.
  
• Whether you're going to the store or to the movie theater, you need to do a bunch of these turns. The code for "go to the store" involves a bunch of identical sequences of code. Not repeating the sequences, but rather keeping one (or actually several, but fewer than one for every needed place!), and then jumping to it whenever you need it is far more efficient (and gets selected for!).
  
• You now have a set of instructions for "going to the store" that have a short sequence of "calls" to these more complex and lower-level instructions on how to turn right and left.
  
• A change in just *one* instruction in that sequence, will land you at the movie theater rather than the store.
  

Ergo, yeah, small changes can have very large effects. This is not built into your assumptions in your math, so you keep coming around to unbelievably improbability in your conclusions, for occurrences that I posit are impossible to avoid rather than being improbable.

 

A similar question which addresses this "modular" approach is to consider whether there is actually enough data represented in a set of genes to produce a complex being *without* the code being modular....something to think about! :doh:

 

Probably exists, :)

Buffy

[Biochemist]

I don't completely agree with this point of view, except in an abstract context.

And I am not suggesting we should conclude that there is unexpressed information load either. It is just that the implication is there, and the presumption of "mutation" whenever the genome changes is, I think, poorly supported.

Each letter in our alphabet has an established meaning. String them together to make words with established meanings. ...Does every word have a latent information load?

Certainly every word does not have latent information load. But the genome is orders of magnitude more complex than the English language, and deserves some special attention just because of the incredible complexity concurrent with incredible efficiency.

 

The human gene count (although still in contention as I understand it) is maybe 30-35,000 genes. The protein count (the "proteome") in any one cell is maybe three times that, due mainly to alternative splicing of the gene by the mRNA. 30,000 genes and 90,000 proteins? The space shuttle has maybe 250,000 parts. Last I saw, it couldn't replicate itself. This suggests that every gene is highly utilitarian.

 

The notion that you can slightly modify one gene and get a significant change in form or function in multiple body areas (usually called pleiotropy) is a significant additional layer of complexity. Not only can we build a very complex organism with a very small number of mechanical parts, but we can advance the apparent complexity orders of magnitude with what appears to be trivial changes.

 

I just think it is presumptuous to assume these trivial changes are actually "mutations".

[Biochemist]

...Ergo, yeah, small changes can have very large effects. This is not built into your assumptions in your math, so you keep coming around to unbelievably improbability in your conclusions, for occurrences that I posit are impossible to avoid rather than being improbable.

Hey Buff. Good to have you on the thread. I figured you would show here.

 

I do understand and agree that modularity is required. I just think there is more than that at work here. 30,000 genes and 90,000 proteins runs an entire cell, including feedback loops to keep protein levels at homeostasis???

 

I am sure we have lots of examples of computer code with more than 30,000 routines. None run as cleanly as a cell.

 

Further, modularity does nothing to mitigate the probability of the advent of a new enzyme system that did not previously exist (the original question in post #1). It is certainly true that we could cut sections of functional genetic code from other portions of the genome. We have documented several mechanisms for that. But we still have the probabilistic issue that the new code has to function to be selected, and it has to first be built to function. The preexisting modules would do nothing to select for interim genomic states.

 

Further, the number of genes (or the respective proteome size) does not map particularly well to the complexity of a species. If you were to slog through this review article on gene count and proteome count:

 

http://nar.oxfordjournals.org/cgi/content/full/30/5/1083

 

you will see that nematodes have about 20,000 genes and flies have about 15,000. Flies seem more complicated (more cells, more functions) and yet have less genetic material. Keep in mind that humans have only 30,000.

 

Overall, in spite of the facts that:

 

• small changes to the genome may result in significant (and dramatic) morpohological change, and
  
• we have multiple mechanisms for duplication of genetic code within the genome,

it still apears to be true that complex biochemical systems arrive on the evolutionary scene suddenly without selection.

 

So stitching all this together,

 

• It is intriguing that we can have such a small number of genes managing such immense biochemical complexity, and then still have the occasional trivial genomic change result in significant morphological change
  
• Nevertheless, there are cases (this subject of this thread) where it appears that entire new genes surface without the opportunity for selection.

Sorry I got so long winded.

[Buffy]

I do understand and agree that modularity is required. I just think there is more than that at work here. 30,000 genes and 90,000 proteins runs an entire cell, including feedback loops to keep protein levels at homeostasis???

Pretty cool eh? I think you're skipping over the whole fact that all those other gizmos running around the nucleus--all the different RNAs, enzymes, proteins, etc--not to mention the fact that the folding of those proteins (not just the sequences) all are part of the information that makes it work. This is something that the biologists have only recently really started to investigate after years of the thinking "well, everything is defined by the gene sequences, why bother looking at the rest of the zoo."

 

More importantly, all these other gizmos are built to function in certain ways that ensure/preclude much of the "pure random permutation" demand of your math: they work in a specific way, so the changes are limited to those that "might" work.

I am sure we have lots of examples of computer code with more than 30,000 routines. None run as cleanly as a cell.

Heh, heh, you betcha! But we've only been writing them for about 50 or 60 years, and as alexander will point out, most of that was without Python, so its all defective code anyway... With Lisp of course you can get self-modifying code with feedback loops in about 100 lines....

 

Actually, its important to note that self-modifying/evolutionary code gets quite ugly, with lots of "failed junk" lying around...just like you see in a cell... ("Is this one of your units, Creator? It is inefficient and fragile and lacks basic protection and mechanisms for self-repair." "It serves me as is, you will repair it" :) )

Further, modularity does nothing to mitigate the probability of the advent of a new enzyme system that did not previously exist ....

Just wanted to repeat this snipet because I'm saying that the non-gene machinery does indeed control/direct the development of these new enzymes, and in almost every case the "new" stuff is indeed modification (maybe radical mods, but mods nonetheless) of an older "selected" sequence...

Further, the number of genes (or the respective proteome size) does not map particularly well to the complexity of a species.

I'm trying to find the ref in a SciAm of the last year, but I've seen something that sez that this is exactly what's getting folks back to look at the non-gene zoo, because it looks like the mechanisms are more complex so there's less need for genes in the "higher" organisms, which would be in line with my argument about how modularization makes for more complexity and higher organisms would have more (and prolly many levels of) modularity...

it still apears to be true that complex biochemical systems arrive on the evolutionary scene suddenly without selection.

And there's still no reason to *assume* that bits of each modification weren't expressed and possibly found "not detrimental" and recessed between environmental shocks, thus creating a "storehouse" of possible changes for when a nasty meteorite hits or CO2 levels go through the roof....

Sorry I got so long winded.

Oh its cute when you're long winded! :doh:

 

Cheers!

Buffy

[Biochemist]

...Actually, its important to note that self-modifying/evolutionary code gets quite ugly, with lots of "failed junk" lying around...just like you see in a cell...

Not really. Although there is indeed lots of non-coding DNA sitting around, it does not mean it is not critical. Unlike self-evolutionary code, we have incredibly efficient machinery.

[Biochemist]

More importantly, all these other gizmos are built to function in certain ways that ensure/preclude much of the "pure random permutation" demand of your math: they work in a specific way, so the changes are limited to those that "might" work.

And the game here is to propose a mechanism, rather than to suggest that "a miracule occurs" and then genes show up.

 

Again, the question is:

 

How do we get a new gene sequence that codes 5 average enzymes when it is not selected before it is expressed?