Waiting for multiple mutations: Intelligent Design Creationism v. population genetics

Casey Luskin is worried about university students. Apparently, they aren't getting enough correct information about intelligent design. Luskin uses the example of a student named Michael Heckle at Iowa State University. Mr. Heckel said; "So far, there has been no research done by intelligent design advocates that has led to any sort of scientific discovery."

This happens to be a true statement but Casey Luskin takes exception in a blog post that appeared the other day on Evolution News & Views (sic) [No ID Research? Let's Help Out This Iowa State Student].

You have to deny mountains of research and evidence to say that. Intelligent design advocates have done a great deal of research, leading to numerous scientific discoveries. Let's help out this student by reviewing some prominent ones, amounting to only a portion of that overall research.

This is going to be fun. Casey Luskin is about to reveal some ID research that has led to some sort of "scientific discovery."¹

The first example that Casey Luskin quotes is Doug Axe's "discovery" that new protein folds cannot evolve. I'm not going to discuss that result. It's pretty obvious that if it were a new scientific discovery then half of my department would be idiots and Doug Axe would be in line for a Nobel Prize.

It's the second example I want to look at ....

In 2004, biochemist Michael Behe and physicist David Snoke published research in the journal Protein Science. They reported the results of computer simulations of the evolution of protein-protein interactions. Vital to virtually all cellular processes, these interactions require a specific "hand in glove" fit, where multiple amino acids must be properly ordered to allow the three-dimensional connection. The simulations showed that the Darwinian evolution of a simple bond between two proteins would be highly unlikely to arise in populations of multicellular organisms if it required two or more mutations to function. They concluded that "the mechanism of gene duplication and point mutation alone would be ineffective...because few multicellular species reach the required population sizes."

Let's look at the paper by Behe and Snoke (2004). (David Snoke is a professor of physics at the University of Pittsburgh.) It's not exactly what Casey Luskin describes. In fact it has nothing to do with protein-protein interactions.

Behe & Snoke looked at the evolution of a new protein function following a gene duplication event. They wanted to model two competing processes. In one situation the "death" of one of the duplicates occurs when it acquires a deleterious mutation. (Shown as a red "x" in the figure.) The probability of such an event is the product of the mutation rate (v ) and the fraction of the mutations that will be deleterious over the entire sequence of the gene (p).

In the other situation, a new gene with a new function is born. In the simplest model they imagine that this "birth" requires three specific mutations in various parts of the gene. The probability is determined by the product of the mutation rate and the number of sites that need to be changed (λ). The example in the figure shows a situation where three amino acid residues have to be changed (green) (λ=3). (The loci could be anywhere in the sequence ... they don't have to be adjacent as shown in the figure.)

The goal is to find out whether the evolution of a "mutliresidue" (MR) feature is possible given standard mutation rates and population sizes. Here's how Behe & Snoke introduce the paper:

Although many scientists assume that Darwinian processes account for the evolution of complex biochemical systems, we are skeptical. Thus, rather than simply assuming the general efficacy of random mutation and selection, we want to examine, to the extent possible, which changes are reasonable to expect from a Darwinian process and which are not. We think the most tractable place to begin is with questions of protein structure. Our approach is to examine pathways that are currently considered to be likely routes of evolutionary development and see what types of changes Darwinian processes may be expected to promote along a particular pathway.

The analysis begins by assuming that the population consists of individuals with a recently duplicated gene. They assume haploid organisms reproducing asexually with no recombination.

The model is going to examine the competition between mutations that "kill" the gene and a combination of mutations that create a gene with a new function. Each of the mutations that will collectively give rise to a new function will disrupt the original function ... it's only when all of the mutations are present that a new gene is born. We will see that this is the assumption that causes the most problems for the model.

The pertinent feature of the model is that multiple changes are required in the gene before the new, selectable feature appears. Changes in these nucleotide positions are assumed to be individually disruptive of the original function of the protein but are assumed either to enhance the original function or to confer a new function once all are in the compatible state. Thus, the mutations would be strongly selected against in an unduplicated gene, because its function would be disrupted and no duplicate would be available to back up the function.

Keep in mind that their model only looks at situations where just one combination of substitutions will work. The parameter, λ, can be 2, 3, 4, or more. It is the total number of different sites that must be changed. These are the "compatible" sites.

The mutation rates of all mutations are the same. One of the important parameters is p—the ratio of deleterious mutations and compatible mutations. If there are 2400 possible deleterious mutations and three compatible sites then p = 2400/3 = 800. (Neutral mutations are ignored.)

The model estimates the average time (T_f) to the first occurrence of the genes with the new function. T_f is the number of generations until the last of the required mutations occur.

The new combination of mutations (= multiresidue feature (MR)) will give rise to a new function with a selection coefficient of s. The new allele will become fixed with a probability of about 2s so the complete equation is:

Behe & Snoke assume that the mutation rate is 10^-8 (v=10^-8). The actual per nucleotide mutation rate is 10^-10 per replication so their value will be off by two orders of magnitude in single-cell organisms but closer to the actual per generation value for large multicellular organisms.

They assume that the ratio of deleterious substitutions to compatible substitutions is 1000 (p=1000). (We'll see in my next post that this value is too high by several orders of magnitude.)

They assume s=0.01. This is a reasonable assumption.

Behe & Snoke calculate the average time to fixation (in generations) for several values of λ and several different population sizes (N). The results are shown in Figure 6 of their paper.

I've added red lines to indicate how we should interpret this figure. For the situation where three different mutations are required (λ=3) for the new function, we can estimate the population size necessary for fixation after 100 million generations (T_fx=10⁸). That value is about 10¹¹ (N~10¹¹).

If the generation time is one year then this means that in order to fix the new gene in 100 million years it would require a population size of 100 billion organisms. This is very unlikely. Behe & Snoke's main point is that the evolution of any new function requires multiple mutations and the probability of these MR features arising in a reasonable time is far too low to be effective—unless you postulate huge population sizes that are far greater than anything we know for most species.

This is an anti-evolution argument. They do not suggest a solution.

The authors admit that some of their estimates might be wrong. They have assumed a population that has already fixed a duplicated gene locus with both copies functional. They claim that their final values of T_fx are underestimates because they should be taking into account the time it takes for the duplicated genes to reach appreciable frequency in the population.

On the other hand, they have assumed that there is only one pathway to a new gene function but there might actually be several different pathways involving a number of different amino acid residues. For example, the new function might require a disulfide bond. Behe & Stoke assume that there are only two possible residues that could mutate to cysteine in order to form such a bond (λ=2 in this case). However, there may be several different residues that could mutate to cysteine so their calculations may exaggerate the difficulty in forming a new function.

They recognize that their value of p (p=1000) might be too high but they claim that it is "conservative" relative to the values used by others. They recognize that the model is sensitive to the value of λ but they quote several examples where new functions require three or more different mutations. Furthermore, even if only two mutations were required it still takes a population size of 10 million to fix the mutations in 100 million years and this is beyond the edge of evolution for most populations.

If Behe & Snoke are correct then this is an amazing result. It means that modern evolutionary theory cannot explain the origins of any new function that requires three or more independent mutations.

But Behe & Snoke are not correct (surprise!). The paper was thoroughly refuted and discredited by Michael Lynch in an article published in the same journal one year later [see the Wikipedia article at: Behe and Snoke, 2004].

In my next post I'll discuss the Lynch paper and Behe's response.

---

1. The most important scientific discovery we learn from ID research is that ID proponents are stupid and/or incompetent. I don't think that's what Casey Luskin means.

Behe, M.J., and Snoke, D.W. (2004) Simulating evolution by gene duplication of protein features that require multiple amino acid residues. Protein science, 13:2651-2664. [doi: 10.1110/ps.04802904]