The Sound of Circular Reasoning Exploding
| December 7, 2006 | Posted by Dave S. under Intelligent Design |
Alternate Title: Of Mice and Men and Evolutionary Dogma
“There has been a circular argument that if it’s conserved it has activity.” Edward Rubin, PhD, Senior Scientist, Genomics Division Director, Lawrence Berkeley National Laboratory
Recent experiments cause a central tenet of NDE to miss the prediction. Large swaths of junk DNA (non-coding, no known function) were found to be highly conserved between mice and men. A central tenet of NDE is that unexpressed (unused) genomic information is subject to relatively rapid corruption from chance mutations. If it’s unused it won’t do any harm if it mutates into oblivion. If it’s unused long enough it gets peppered with mutations into random oblivion. If mice and men had a common ancestor many millions of years ago and they still have highly conserved DNA in common, the story follows that all the conserved DNA must have an important survival value.
A good experiment to figure out what unknown purpose the non-coding conserved pieces are doing would be to cut them out of the mouse genome and see what kind of damage it does to the mouse. So it was done. Big pieces of junk DNA with a thousand highly conserved regions common between mice and men was chopped out of the mouse. In amazement the mouse was as healthy as a horse (so to speak). The amazed researchers were in such a state because they were confident NDE predicted some kind of survival critical function and none was found.
This is a good avenue for positive ID research. If the function of any of those regions were preserved because they could be of important use in the future… well that would pretty much blow a hole in the good ship NDE the size of the one that sunk the Titanic. Maybe not that big, but it would be taking on water – natural selection can’t plan for the future. Planning for the future with genomic information is the central tenet of ID front loading hypothesis. Lack of any known means of conserving non-critical genetic information is the major objection lobbed at the front loading hypothesis. Evidently there is a means after all.
Life goes on without ‘vital’ DNA
16:30 03 June 2004
Exclusive from New Scientist Print Edition.
Sylvia Pagán Westphal, BostonTo find out the function of some of these highly conserved non-protein-coding regions in mammals, Edward Rubin’s team at the Lawrence Berkeley National Laboratory in California deleted two huge regions of junk DNA from mice containing nearly 1000 highly conserved sequences shared between human and mice.
One of the chunks was 1.6 million DNA bases long, the other one was over 800,000 bases long. The researchers expected the mice to exhibit various problems as a result of the deletions.
Yet the mice were virtually indistinguishable from normal mice in every characteristic they measured, including growth, metabolic functions, lifespan and overall development. “We were quite amazed,” says Rubin, who presented the findings at a recent meeting of the Cold Spring Harbor Laboratory in New York.
He thinks it is pretty clear that these sequences have no major role in growth and development. “There has been a circular argument that if it’s conserved it has activity.”
Use the link above for the full article.
134 Responses to The Sound of Circular Reasoning Exploding
Leave a Reply
You must be logged in to post a comment.
DaveScot:
I think it is simply a failed prediction of NDE. I know there was a strong prediction that generation time would correlate with the mutation rate. That has not turned out to be the case.
If NDE were true I would expect to see uniform drift across the genome with a close correlation to generation times, exceptions would only be made for genes under selective pressure that cannot tolerate neutral mutations as well. The Nature Mouse Genome issue made this comment:
Sparc:
Jehu’s quote from Nature:
What Jehu alludes to, and what Sparc’s quote actually points out, is the fact that irrespective of the question concerning conserved sequences and function there seems to be TWO conserving processes going on, one that can be plausibly attributed to NS, and one that defies this connection. But there’s more to it than that. EVEN IF NS is responsible for both such processes (thus, granting the argument), the “enhancers” (per Pennacchio) are nevertheless more conserved than the “genes” themselves. These enhancers “regulate” gene function/expression. Hence, it would seem–using the standard thesis regarding conservation of sequences–that ‘regulated’ functions are more critical to survival than are “coding” functions( i.e., proteins/enzymes).
Dave, I’m going to toss this over to your area of expertise–programming–but doesn’t this all suggest that “genes” are less critical in, so to speak, setting up genetic programming than are the “enhancers/regulators” themselves? E.g., wouldn’t a problem with a branching node be more fatal to the proper functioning of a program than a called for subroutine? IOW, in programming you’d fix the switching problems first before you started looking at the individual subroutines the branching node was calling for. Looked at this way, it makes gene expression look somewhat secondary, almost peripheral, to genetic programming–which is really what we see in nature, as in the geographical radiation of species. And doesn’t that imply that NS is necessarily almost decoupled from phenotypic variation? And doesn’t that imply that Darwin was completely wrong given that he bases his theory on the link between phenotypic variation and selection?
DaveScot:
In true “junk dna” mutations randomness determines when a mutation becomes fixed. However, if we look at the amount of benefit that a mutation must offer so that it does become fixed, we can guess that the converse is also true, that a mutation that offers that much disadvantage would keep it from becoming fixed. If DNA is actually ultra-sensitive to any advantage/disadvantage, then any mutation offering even a very slight disadvantage would not be fixed. The result would be that that segment of DNA would be conserved.
Mesk, over on Telic Thoughts, argues that it is quite reasonable for the mice in question to pass the tests provided and still have sufficient deleterious function account for the conservation that is observed. However, he suggests that a 5 year study on about 1000 mice, may make a very convincing case that the mice have suffered no deleterious effects. He then goes on to describe the expected response from the scientific community (assuming that no deleterious effects are found) as frantic. I do like the balance the Mesk is offering in his discussion on TT.
I’ve been thinking of a simple simulation that would rule out the “segments might just have not been mutated” argument. (If I can find a few hours I will code it.)
The code would create an million (tuneable) element array, and randomly mutate its values between a value of 1 and of 4. Let sufficient mutations happened to account for random mutations over the years allotted (2.2% per million years * 140 million years, right?) . At that point, the array can be swept to see what the longest segment that was apparently unmutated to the 70th percentile would be. I bet the longest segment will be about 30 elements long. I will be shocked if the array will show a thousand segments averaging 100 element in length — shocked, shocked!
DaveScot
Although I am guilty of giving that number earlier in the thread, it is apparently actually much lower. According to Nature the mutation rate in the mouse is 4 x 10^-9 per year per base pair and in the human is 2 x 10^-9 per year per base pair. So those two mutation rates have to be combined. The time of divergence for the human/mouse divergence is 70 – 90 million years, which resulted in neutral divergence of 47%.
The exons of coding genes in mouse to human have an average of 85% identity. The introns average 69%. The so-called “promoter regions” which are the poorly defined 200 base pairs just before and after a coding gene have an identity of between 70-75%.
I am not sure how significant 70% identity between mouse and human is. However, when you toss in a chicken in there the time of divergence goes way back and it is much harder to justify it by random chance.
That would be bFast “…. right?”
That right, I meant bFast. Sorry.
Jehu, let me question your figures. We seem to have two sets of numbers flying around. Could the 2 x 10^-9 be an “average rate for all DNA” where the 2.1% per mil be the rate for “junk” DNA? I have independantly seen the latter number measured off of “junk” DNA, so that’s what I am suspecting.
bFast
2.22 x 10^-9 is the average mutation for mammals, I am not sure how accurate that number is because it is from 2001. The number’s I have been giving for the neutral substitution are from Nature‘s mouse genome article. (It’s free) http://www.nature.com/nature/j.....01262.html
Here’s how to calculate the significance of the conservation. Assume that by whatever figures you want to use, you expect by chance a certain percentage of bases to have changed, so like 47% divergence (53% identity), whatever you’d like to use.
Now, what is the probability that a 100-bp window will be preserved at 70% identity? You have 100 bases and fewer than 30 of them have to have changed, where each one has a 47% chance to have diverged. If I flip a coin a hundred times, and it has a 47% chance of coming up tails, what’s the probability I get fewer than 30 tails? You compute this using the cumulative binomial distribution. There is not a simple formula, but you can go plug it in to Matlab, etc. At 47% divergence, the chance of this happening for a 100-bp window is 1.8e-4.
How many 100-bp windows are there in the genome? A lower bound is 3e9/100, i.e. let’s not even let the windows overlap. So how many 100-bp windows in the genome would we expect to be preserved at better than 70% identity by chance? 3e9/100*1.8e-4 = 5,400. At 40% divergence, it’s 83,000. Again, both lower bounds because we considered only disjoint windows.
May I once again stress that I was never arguing that many conserved sequences were not deleted in the Nobrega experiment; this calculation arose in a side discussion. But since my math was being questioned, I thought I’d bring the bacon.
Finally, let me caution against the calculation several of you have been doing (and, for simplicity’s sake, I also used in a previous comment) of X subs/site/yr times Y years to get a percent divergence. This calculation becomes increasingly inaccurate for a larger number of years. The mutation rate is of the process where you randomly pick a position in the genome and change it to another base. The trouble with trying to get percent identity/divergence from this is that as you let this process run, on each iteration you have a higher chance of changing a base that you already changed, leading to no decrease (and possibly an increase) in the percent identity. Percent identity is a nonlinear function of the time passed, and you need a differential equation to model it. I am not sure off the top of my head where (back in time) the linear approximation that we’ve been using becomes really bad.
GeoMor,
Thanks for input. But help me out here. The mouse genome is 2.5 gb. So your equation should be 2.5e9/100*1.8e-4=4,500. So we have 4,500 100bp sequences that would have 100 bp and 70% homology or .00015% of the genome? So there will be a small amount of falsely conserved sequences. Rubin’s team found 1,243 with >70% homology and 100 bp in only two stretches of DNA between four exons. I take it there are more of these sequences than would be anticipated by the normal curve.
Not if you use the 47% figure because that is based on the hard observation of of comparing the human and mouse genome.
Things get real interesting when you also consider that Rubin’s team could not find function for 5 sequences of >90% identity and 180 bp or for only 1 of 10 sequences of >90% idenity and 400 bp that were conserved across humans, rodents, chickens and frogs. There is no evidence that the function they did find conferred a selective advantage.
Correction to that last post. Rubin’s team only found 1 function out of 10 sequences of >90% idenity and 400 bp that were conserved across humans, rodents, chickens and frogs. There is no evidence that the function they did find conferred a selective advantage
GeoMar: “Percent identity is a nonlinear function of the time passed, and you need a differential equation to model it.”
Mathemetitions, puh!
If you set up the array that I described, it will automatically bump into mutated items, and mutate ‘em again. Wala, advanced differential calculus.
You can drive yourself crazy trying to calculate the trajectory of a football in a stiff breaze, or you can ask a quarterback to throw the darn thing.
[...] This is a good avenue for positive ID research. Planning for the future with genomic information is a central tenet of the ID front loading hypothesis. Lack of any known means of conserving unexpressed genetic information is the major objection lobbed at the front loading hypothesis. Natural selection is the only mechanism known for preserving genomic information and to do it the information must be “expressed” so that it has some testable survival value for selection to act upon. If it’s not expressed then it is subject to eventual destruction by natural selection’s ever present companion “random mutation”. Evidently there is a means of preserving unexpressed information after all. See also this related blog article I wrote two months ago which is even stronger evidence of a genetic information cold-storage mechanism: The Sound of Circular Reasoning Exploding. Rogue weeds defy rules of genetics 00:01 23 March 2005 NewScientist.com news service Andy Coghlan [...]