Is There an Edge to Evolution? Part 5: It’s All About Numbers

Bookmark and Share

November 19, 2010 Tags: Design

Today's entry was written by David Ussery. Please note the views expressed here are those of the author, not necessarily of The BioLogos Foundation. You can read more about what we believe here.

Is There an Edge to Evolution? Part 5: It’s All About Numbers

Dr. Ussery continues his chapter-by-chapter analysis by focusing on Chapter 8. This time he finds some significant problems with Behe’s extrapolations. Darrel Falk and Dave Ussery have worked closely on this; however, the primary author is Dr. Ussery.

Chapter 8 - Objections to the Edge

I agree with Behe when he says “Time is actually not the chief factor in evolution - population numbers are.” (page 153). Perhaps an analogy can help explain this. In my line of work, we rely heavily on computers. For example, I want to do a comparison of a thousand bacterial genome sequences – if it takes a few days to do a calculation on one genome, then it would take literally YEARS to do the calculations for a thousand genomes. How do we get around this? By using lots of processors in parallel. If we have 1000 CPUs, then in principle, assuming the computers are free and all goes well, we can do the calculation in a few days. Thus, by using parallel processing, one can speed things up tremendously. The argument goes for evolution as well. Although the mutational frequency might be small, if you have enough genomes, the chances of getting the ‘right combination’ is much greater, especially if it happens in parallel along with the occasional recombination of genomes.

Behe’s argument in this chapter is essentially that even with more than several hundred million years of evolution, this is simply not enough time for the ‘right mutations’ to occur in order for the complexity we see around us, in terms of plants and animals, to have evolved via ‘random processes’. On page 163, Behe poses the question: "Yet if it can do so little, why is random mutation/natural selection so highly regarded by biologists?" He then goes on to compare the idea of random mutations with that of "ether", that mysterious substance hypothesized to exist more than a hundred years ago, but thoroughly discredited by Einstein. It is quite clear from this comparison that Behe thinks “random mutation” is a myth believed by most biologists on faith, with little evidence to back it up.

I disagree. I do believe that life’s history is infused with purpose and that this process is God’s process. The question here, from my perspective, is not whether there is purpose or not, but whether the scientific arguments presented in Behe's book make sense and are valid, based on what is currently known in biochemistry and molecular biology. It is those arguments that I address here. To really understand the potential of mutations to build new protein interactions you need to see a much bigger picture than Behe paints. Bacteria have been around since the first ecosystems, more than 4 billion years ago, and are still the most predominant life form on the planet today. I have a table I love to show my students when I'm teaching. It comes from a review article published about a year ago. There are 1031 bacteriophages (viruses that attack bacteria) on the earth, and if one were to stretch out their genomes, end-to-end, they would be about a thousand times the length of the Milky Way galaxy! If one were to stretch out all of the bacterial DNA from the planet, it would be close to a MILLION times the length of the Milky Way! So this is an enormous amount of DNA. Since bacteria have very short lifetimes (less than a day) that means that more than that amount of DNA is being replicated every day. With each replication there is an opportunity for genetic change in parallel lines which have the opportunity to mix and match every so often in the history of life. In examining a tiny, tiny fraction of that, a 'mere' thousand bacterial genomes, I am absolutely astounded at the amazing diversity. As I've said before, not a single protein is conserved amongst just this tiny sampling of bacteria we've looked at so far, and many bacterial 'species' have less than half the proteins of one genome found in another genome - of the same species! To what extent does Behe appreciate this vast opportunity to build new combinations of proteins?

Behe makes an astonishing conclusion. He states “the formation of even one helpful intracellular protein-protein binding site may be unattainable by random mutation.” (page 157). Let’s start off by examining what has been published. Go to PubMed, search their more the 20,000,000 articles online. If you type in “evolution, protein binding sites” you will see the article, “Structural features and evolution of protein-protein interactions” along with 5400 other articles on the topic. The abstract for this article includes the sentence:

Here, the interfaces of 750 transient protein-protein interactions as well as 2,000 interactions between domains of the same protein chain (obligate interactions) were analyzed to obtain a better understanding of molecular recognition and to identify features applicable for protein binding site prediction.

This is just one article. Would you agree that perhaps Behe’s statement “the formation of even one helpful intracellular protein-protein binding site may be unattainable by random mutation” is likely not to be too meaningful? It seems that it might be a little premature to bring his summary of the state of biological research to a public audience as he did in this book. There is no question that Behe’s story is very incomplete. You are especially urged to read Kelsey Luoma’s excellent article on this. She is an undergraduate student who did what all good science students do--she went back to check the literature. The literature clearly demonstrates the evolution of new protein interactions.

So Behe is clearly wrong when, on page 154, he says that since “we see no new protein-protein interactions developing in 1020 cells, we can be reasonably confident that, at least, no new cellular systems needing two new protein-protein interactions would develop in 1040 cells - in the entire history of life…" Depending upon your math background you might be tempted to think that the difference between 1020 and 1040 is not that great. Just in case that is the case, let’s examine how different those numbers are with a little illustration. The DNA from 1020 cells of bacteria would be about 18 light years long – that’s a lot of DNA! However, the length of the DNA from all bacteria, on the face of the planet, living right now (roughly 1031 cells), is about 100,000,000,000 LIGHT YEARS long. However, that is just is just the amount of bacterial DNA present right now. Bacteria duplicate as often as once every five minutes. So compared to the DNA in 1020 cells (18 light years) the amount of DNA in 1040 cells is 1,800,000,000,000,000,000,000 light years. That’s a lot of DNA. (Remember there are 180,000 miles in one second of a light year. That’s a lot of DNA.) Let’s be careful about telling the public “we can be reasonably confident that, at least, no new cellular systems needing two new protein-protein interactions would develop in 1040 cells - in the entire history of life…” The generation of this amount of DNA provides for a lot of opportunity for mutations that would generate new protein interactions.

Let’s look further at what really was done in the experiment with 1020 cells he discusses in the quote from page 154 where he clearly states that no new protein-protein interactions were seen. The fact is that in this experiment they didn’t search the proteome for new protein-protein interactions - they were only looking for one particular type of mutation. So not only did Behe’s extrapolate from a “pin-prick” sample size (1020 cells) to a larger than universe-sized sample size (by comparison), the authors of this study didn’t even begin exhaustively comb the “pin-prick” sample for new protein-protein interactions. It is dangerous to extrapolate over “zillions” of orders of magnitude (from 1020 to 1040) even at the best of times. However, Dr. Behe did it for a parameter that had not even been carefully searched to begin with. The investigators did not design the experiment to search for any new protein-protein interactions in the entire protein repertoire of cells- they were just probing for one particular phenotype. Behe is correct that they didn’t see them, but to conclude that they didn’t find ANY new protein-protein interactions is a bit far-fetched, since they weren't looking for them. They were only looking for a small number of highly specific changes, not the proteome as a whole. True, no one reported finding beneficial mutations in the samples studied, for this particular case, but to conclude that they can in general never or only rarely happen is just a hopeful extrapolation.


David Ussery is an associate professor of comparative microbial genomics at the Center for Biological Sequence Analysis at the Technical University of Denmark and on the faculty at the University in Oslo, Norway. Ussery is the co-author of Computing for Comparative Microbial Genomics and has authored or co-authored 130 articles for science and professional journals. He is also a frequent public speaker on the topic of bacterial genomics.

< Previous post in series Next post in series >


Share your thoughts

Have a comment or question for the author? We'd love to hear from you.

View the archived discussion of this post

This article is now closed for new comments. The archived comments are shown below.

Loading...
Page 7 of 9   « 4 5 6 7 8 9 »
Alan Fox - #42511

December 4th 2010

I’m sorry but not surprised that you don’t like the answer, chunk. Try asking some experts elsewhere; Telic Thoughts maybe. There must be someone there that will supply the answers you claim to seek.


chunkdz - #42513

December 4th 2010

Alan:

I’m sorry but not surprised that you don’t like the answer, chunk.

It’s true - I don’t like your answer. Not because you say “of course MET predicts this”, but because you’ve failed to understand the question. The reason I can tell that you don’t understand the question is because when I ask you what percentage of homologs we expect to find you refuse to answer.

Try again. Do we expect to find 10%? 50%? 90%? 100%? 75%? 30%? .01%?

If it’s such a “basic” question and if it is such a “basic” prediction of MET then you should “basically” be able to make a “basic” falsifiable prediction and include a “basic” rationale to support your “basic” prediction so that we may “basically” falsify your “basic” prediction experimentally.


Alan Fox - #42514

December 4th 2010

@ chunk #42513

But that’s a different question to:

Just curious - is this a prediction that flows from Modern Evolutionary Theory…

I would need to know how you define “homologue/homolog” before giving a precise answer. I guess you won’t like what Wikipedia has to say on homologous sequences:

The phrase “percent homology” is sometimes used but is incorrect. “Percent identity” or “percent similarity” should be used to quantify the similarity between the biomolecule sequences. For two naturally occurring sequences, percent identity is a factual measurement, whereas homology is a hypothesis supported by evidence.

What exactly do you mean by “percentage of homologs”?


Alan Fox - #42515

December 4th 2010

BTW when you say “...we may “basically” falsify your “basic” prediction experimentally.” do you have an associate with lab facilities or are you apeing Zachriel?


chunkdz - #42519

December 4th 2010

Alan:

But that’s a different question to:

Not really. Dr. Ussery said “many (most?)”. Pending further clarification I judge this to be somewhere above the middle range of percentages. 

I would need to know how you define “homologue/homolog” before giving a precise answer. I guess you won’t like what Wikipedia has to say on homologous sequences:

Percent homology is NOT “percentage of eukaryotic proteins that have prokaryotic homologs.”


Like I’ve been saying - I suspect you don’t understand the question, or how to determine the answer. You won’t find it in Wikipedia, I’m afraid.


Dr. Ussery is a bioinformatics expert. He knows precisely how to interpret BLAST results, he knows how to infer homology, and his lab is well equipped to make predictive assessments about homology.

What do you say we wait patiently for the good professor to answer the question?


John - #42554

December 5th 2010

Chunkdz wrote:
“But as my question was not about Doug Axe or his hypothesis, I’d prefer to wait and let Dr. Ussery answer my question. I hope you understand.”

I wasn’t addressing your question. I was challenging your shallow and hypocritical claim about text trumping data:
“Actually you linked to the Szostak paper as putative evidence that novel proteins evolve from extant proteins. Unfortunately, that Szostak paper says nothing of the sort.”


Alan Fox - #42669

December 6th 2010

Chunk:

Dr Ussery comments in the newer thread (in a response to Bilbo):

...ummm, I don’t get it.  How can proteins NOT be ‘borrowed from a previous system’ - unless they are created de novo by God or something.  I really don’t understand this question.  Can you clarify this a bit and give me an example of what it is that you are asking?

Which indicates to me, as all organisms (if evolution is true) descend with modification from a common ancestor, so genes and therefore proteins have the same nesting hierarchies and thus homology is 100%. Of course, as we can only examine DNA and proteins from very recently extinct and living organisms, the evolutionary paths and similarities may no longer be discernible.

You may want to check the new thread..


chunkdz - #42681

December 6th 2010

Alan:

Which indicates to me, as all organisms (if evolution is true) descend with modification from a common ancestor, so genes and therefore proteins have the same nesting hierarchies and thus homology is 100%.

Ok, so Dr. Ussery said “many (most?)” and Alan Fox says 100%.

Alan, if the human gene for catenin does not have a homolog among eubacteria and archaea would you concede that your prediction of 100% might possibly be a bit overstated?


chunkdz - #42682

December 6th 2010

John:

I wasn’t addressing your question. I was challenging your shallow and hypocritical claim about text trumping data

The Szostak paper demonstrates quite nicely that function can appear without any evolution at all.

At any rate it has nothing to do with predicting how many homologs MET expects to find.


John - #42690

December 6th 2010

chunkdz wrote:
“The Szostak paper demonstrates quite nicely that function can appear without any evolution at all.”

and:

“At any rate it has nothing to do with predicting how many homologs MET expects to find.”

It has everything to do with it. You’re beyond help if you can’t see the obvious contradiction between your own statements.


Alan Fox - #42712

December 6th 2010

Alan, if the human gene for catenin does not have a homolog among eubacteria and archaea would you concede that your prediction of 100% might possibly be a bit overstated?

No. Homology is the hypothesis. Take oxytocin. Nine residues. Hypothesis: evolved by single point mutations from random sequence. Able to spot homology over three billion years, unlikely!


chunkdz - #42755

December 6th 2010

Alan:

Able to spot homology over three billion years, unlikely!

Dr. Ussery doesn’t seem to think so. He says many or most eukaryotic proteins should have homologs in bacteria.


Alan, I really think you are not getting the concept. Perhaps if you simply wait for Dr. Ussery to answer the question he might be able to clear up your confusion.


Alan Fox - #42796

December 7th 2010

At large evolutionary distances, e.g., between Archaea and Bacteria, sequence similarities may be eroded to such an extent that the distance between orthologous sequences is similar to that between sequences that are merely part of the same gene family. More dramatically, homolog sequences can diverge “beyond recognition,” such that the similarity between two orthologs is not higher than the similarity between sequences that are not part of the same gene family and automatic procedures for the recognition of homology fail.

Link


chunkdz - #42867

December 7th 2010

This is supposed to be evidence that MET predicts we will find bacterial homologs for most eukaryotic proteins??

You really should take my advice, Alan, and wait for Dr. Ussery’s answer.


Alan Fox - #42987

December 8th 2010

@ chunk

MET predicts homolgy in gene (and thus protein) sequences but “homolog sequences can diverge “beyond recognition,” . That doesn’t prevent a great deal of research which has been and is being done to rationalize the huge amount of available sequence data into nested hierarchies. Not all homologies may be traceable but there is no evidence so far contradicting common descent and masses in support.

BTW, I think Dr Ussery has answered your question:

Dave Ussery wrote: “I wouldn’t be surprised if many (most?) of the eukaryotic proteins have homologs in bacteria - this just makes sense…”

Just curious - is this a prediction that flows from Modern Evolutionary Theory, or is it a novel prediction based upon your own research?

so I am not waiting for any further response.

If you really think your question is deserving of a further response why not post it in the newer thread. This one has obviously run its course.


chunkdz - #43022

December 8th 2010

Alan:

MET predicts homolgy in gene (and thus protein) sequences but “homolog sequences can diverge “beyond recognition,”

Some do. Some don’t. Does MET predict “many (most?)”? In the case of beta catenin we do not find a homolog in eubacteria or archaea. Did MET predict this? What does “many (most?)” mean? 50%? 90%? What is the rationale for predicting “many (most?)”? Is it simply the vast variety of bacterial genomes and therefore a game of sheer numbers? Is it a functional prediction based on bioinformatics?

BTW, I think Dr Ussery has answered your question:

No, not yet.

so I am not waiting for any further response.

You didn’t wait before. I didn’t expect you to wait now. Unfortunately for you, you seem satisfied with your own uninformed answer which is essentially “MET predicts whatever we find”.

If you really think your question is deserving of a further response why not post it in the newer thread. This one has obviously run its course.

I’m sure Dr. Ussery is very busy but will get around to answering the question when he gets a moment. But I agree that your argument has run it’s course.


John - #43037

December 8th 2010

chunkdz:
“Does MET predict “many (most?)”?”

Yes. Does ID?

“In the case of beta catenin we do not find a homolog in eubacteria or archaea. Did MET predict this?”

Specifically? No. Generally and quantitatively, yes.

” What does “many (most?)” mean? 50%? 90%?”

Around 50%, of course.

“What is the rationale for predicting “many (most?)”?”

Evolutionary theory, known genetic mechanisms, and data.

“Is it simply the vast variety of bacterial genomes and therefore a game of sheer numbers?”

Absolutely not. Perhaps you should read up on BLAST and do a few runs, but not in the shallow way that “Mike” did. It’s a statistical analysis.

“Is it a functional prediction based on bioinformatics?”

It’s a prediction of MET based on the evidence. The evidence was massive before anyone coined the term. Have you considered examining the evidence for yourself?


Alan fox - #43048

December 8th 2010

Some do. Some don’t.

So you agree with “homolog sequences can diverge “beyond recognition,””? You understand that if two living species diverged from a common ancestor 3 billion years ago, that’s effectively six billion years of evolution’s worth of possible diversification. Repeating, homology is the prediction of MET. Clear evidence may not be found or exist in every case. We have no sequences from early living organisms, remember.

Echoing John, out of curiosity, allowing for the purpose of argument that MET is a very poor and incomplete theory, do you have a better one?

What is the rationale for predicting “many (most?)”?

Common descent predicts that modification will be stepwise through viable intermediates and form a nested hierarchy. Homology can be demonstrated by comparing sequences across species. The primary papers are full of such comparisons and the resultant nested hierarchies.

I know you to be a supporter of Mike Gene (or at least you give that impression). How does Gene’s “frontloading” conjecture fit the facts or explain anything?


Ashe - #43055

December 8th 2010

Alan:

Repeating, homology is the prediction of MET. Clear evidence may not be found or exist in every case. We have no sequences from early living organisms, remember.

Homology is a prediction of the Modern Synthesis, but not deep homology. Deep homology was unexpected. Biologists Kirschner and Gerhart pointed this out recently:

Deep conservation was unanticipated by evolutionary theory. Mayr, for example, did not think that similarity in structure, such as bat and bird wings, was caused by true homology. He thought that ‘many of the higher categories are unnatural groupings of unrelated animals that have become very similar owing to convergence’. .. This serious misconception speaks to the weakness of the contemporary knowledge of developmental biology at the time of the Modern Synthesis.

Evolution: The Extended Synthesis p. 269

Deep homology is in fact anticipated by Mike Gene’s idea of front-loaded evolution.


Page 7 of 9   « 4 5 6 7 8 9 »