In Behe’s last book, Darwin Devolves, he makes the claim that the authors of a study on polar bear evolution agree with him that:
- Mutations in ApoB are selected and confers benefits to the polar bear. They are phenotypically beneficial.
- These mutations, however, are biochemically damaging to the ApoB.
This is the key claim underlying his Hypothesis of Devolution:
H1 (PB implies BD): Most changes that are phenotypically beneficial (PB) are also biochemically damaging (BD).
If this is false, his argument fails, entirely. From our point of view, even if this is true, he neglects constructive neutral evolution, so it doesn’t really matter. He has a lot more work to do to make his case. Still, if this hypothesis is false, his whole argument is back to square one.
Behe wants to test this hypothesis is by looking at selected mutations (which are phenotypically beneficial), and determining if most of them are “biochemically damaged.” If most of them are biochemically damaged, well, that would support H1. He would still have more work to do, but that would be a good start. Let’s see how H1 fairs on ApoB.
The Original Study’s Authors Said.
First, let us look at what the original authors wrote. They write in their abstract,:
One of the genes showing the strongest evidence of selection, APOB , encodes the primary lipoprotein component of low-density lipoprotein (LDL); functional mutations in APOB may explain how polar bears are able to cope with life-long elevated LDL levels that are associated with high risk of heart disease in humans.
Notice, that they did not call ApoB “damaged.” Why not? There is no reason to think ApoB is damaged. Abstracts are where conclusions are stated, and that is a key conclusion of this paper.
Let us see what they said in the paper. In scientific writing, it is common to list out several observations, and draw a conclusion from them. That is exactly what we see in this key paragraph, with Behe’s quote mine and the authors conclusion bolded.
Due to a lack of appropriate functional studies of polar bears, we were unable to directly identify causal variants. Nevertheless, we assessed the impact of polar bear—specific substitutions on human proteins for top-20 genes under positive selection by computational predictions: a large proportion (ca. 50%) of mutations were predicted to be functionally damaging (Figures 4C and 4D, Table S7). Substantial work has been done on the functional significance of APOB mutations in other mammals. In humans and mice, genetic APOB variants associated with increased levels of apoB are also associated with unusually high plasma concentrations of cholesterol and LDL, which in turn contribute to hypercholesterolemia and heart disease in humans (Benn, 2009, Hegele, 2009). In contrast with brown bear, which has no fixed APOB mutations compared to the giant panda genome, we find nine fixed missense mutations in the polar bear (Figure 5A). Five of the nine cluster within the N-terminal βα1 domain of the APOB gene, although the region comprises only 22% of the protein (binomial test p value = 0.029). This domain encodes the surface region and contains the majority of functional domains for lipid transport. We suggest that the shift to a diet consisting predominantly of fatty acids in polar bears induced adaptive changes in APOB , which enabled the species to cope with high fatty acid intake by contributing to the effective clearance of cholesterol from the blood.
We see this pattern here: a list of observations, followed by a conclusion. Here are the observations in this paragraph:
- lack of appropriate functional studies
- under positive selection by computational predictions
- a large proportion (ca. 50%) of mutations were predicted to be functionally damaging (This is the quote mine from Behe)
- In humans and mice, genetic APOB variants associated with increased levels of apoB are also associated with unusually high plasma concentrations of cholesterol and LDL, which in turn contribute to hypercholesterolemia and heart disease in humans
- brown bear, which has no fixed APOB mutations compared to the giant panda genome, we find nine fixed missense mutations in the polar bear
- Five of the nine cluster within…[the] domain [that] encodes the surface region and contains the majority of functional domains for lipid transport
Now, finally, to their conclusion, which is echoed in the abstract, and qouted several times to Behe (see, for example, here Behe's Trainwreck Response to Science).
- We suggest that the shift to a diet consisting predominantly of fatty acids in polar bears induced adaptive changes in APOB , which enabled the species to cope with high fatty acid intake by contributing to the effective clearance of cholesterol from the blood.
Observations are not conclusions. The authors observe that Polyphen 2 predicts “damaging,” but they do not conclude this is correct. In fact, Lenski, Lents and I have already presented direct evidence of what the authors actually thought, in a follow up news report.
In a news piece about this research, one of the paper’s authors, Rasmus Nielsen, said: “The APOB variant in polar bears must be to do with the transport and storage of cholesterol … Perhaps it makes the process more efficient.” In other words, these mutations may not have damaged the protein at all, but quite possibly improved one of its activities, namely the clearance of cholesterol from the blood of a species that subsists on an extremely high-fat diet.
On damaged genes and polar bears | Telliamed Revisited
Behe is welcome to disagree with these authors, but he cannot claim the authors agree with him. The authors DID NOT conclude that the genes were damaged. They observed this was a prediction of PolyPhen, that is it. Claiming they agree with Behe that ApoB is damaged it is totally and utterly false.
Testing the Hypothesis
Behe disagrees with the original authors, Richard Lenski, Art Hunt, Nathan Lents, and myself, arguing that ApoB is damaged. We have explained that whether ApoB is damaged or not, a “damaged” prediction from Polyphen does not tell us one way or another. A great way to demonstrate this is with a hypothesis test. We will consider a few different hypotheses here.
H1: ApoB is biochemically damaged.
H2: ApoB not biochemically damaged, but functionally improved.
The way how hypothesis testing works is that we now work out what we expect ot see under both hypotheses, then we look at the data.
If H1 is correct, what do we expect? We expect to see EITHER mutations such as truncations, deletions, and frameshifts that are very likely to break the protein OR missense mutations that are predicted to be “damaging” or “benign” by PolyPhen.
If H2 is correct, what do we expect? We expect to see (most likely) missense mutations that are predicted to be “damaging” or “benign” by PolyPhen.
Remember, PolyPhen 2 only predicts “benign,” “possibly damaging,” and “probably damaging.” It does not predict “beneficial” or “constructive” (more on this in a moment). So we expect biochemically constructive mutations to be called “damaging” (or maybe “benign”) and we don’t expect to see any truncations or large deletions (small in-frame deletions are okay).
So, what do we observe?
- Polyphen “damaging” mutations
- Polyphen “benign” mutations
- No truncations, frameshifts, or deletions.
How does this fair with the two hypotheses? Both of them predict 1 and 2, but only H1 predicts the opposite of 3. We only see 1 and 2, so this supports H2, without necessarily ruling out H1, though certainly challenging H1.
This is consistent with both H1 and H2, but H2 is a more precise prediction, so this is data in support of H2. Of course, H2 could be wrong here, but the evidence is tipped in its favor because we do not see any truncations, which could have clearly pointed towards H1. This is not evidence for Behe’s hypothesis at all. This is just the basic rules of hypothesis testing. Behe’s hypothesis (H1) is NOT supported by this data, but challenged by it. The data does not rule out H!, but does not support him at all.
PolyPhen Predicts Phenotypically Damaging
The problem for Behe gets even deeper if we look at what Polyphen 2 means by “damaging.” Does the program mean phenotypically damaging or biochemically damaging? Behe wants it to mean biochemically damaged, but the program means phenotypically damaging.
The data set that PolyPhen is using looks at mutations in humans that cause disease, and tries to pick these disease causing (phenotypically damaging) mutations from normal variation (benign). That is why the term “benign” is used, because the original purpose of the software was to help figure out what are the disease causing mutations in sick patients. Note, therefore,
Polyphen “damaged” means phenotypically damaging, not biochemically damaged, the exact opposite of Behe’s reading.
Polyphen does not attempt to predict “biochemical damage.”
Biochemical damage (as defined by Behe) is idiosyncratic to him, and inconsistently defined, so there are no software that predict Behe’s version of “biochemically damaged,” certainly not PolyPhen 2.
Polyphen is being used outside its original domain when applied to selected mutations, because selected mutations by definition are not phenotypically damaging.
In the case of Polar Bear ApoB, we know the mutations were not phenotypically damaging (they were selected!), which is why most scientists (including the original authors) have correctly interpreted them to be “change/improvement of function” mutations, not “biochemically damaging.”
That is how the contradiction between “selected” and “phenotypically damaging” is resolved, by realizing that Polyphen “damaging” is a proxy for “change of function.” It is not, however, a valid proxy for “biochemically damaging.”
Repeating a Quote Mine, Ignoring the Hypothesis Test
Behe and ENV’s response has been repeating falsehood.
They falsely claim the original authors concluded the mutations were damaging. This is false, demonstrated by (1) the abstract of the paper, (2) the paragraph they selectively quote from, (3) new reports quoting the authors, and (4) repeated explanations of this by Lents, Hunt, Swamidass, and Lenski.
They falsely claim that polyphen 2 predicts the mutations are biochemically damaged. This claim is false, as PolyPhen 2 predicts phenotypically damaging, not biochemical damage.
What is fairly remarkable to scientists watching this is that we don’t test ideas with quote mines. All of us would drop the clearly false quote mine, and go straight to the data. We are hoping that Behe can give that a shot.
The data gives us a way to test his hypothesis.
As I laid out here, it is consistent with his hypothesis, but supports the alternate hypothesis, that these were biochemically constructive mutations. Of course, we need to do actual biochemical studies to know for sure. However “biochemically damaging” is very poorly defined for missense mutations. It is unclear what this could possibly mean. The fact that none of these were truncation mutations counts against Behe’s hypothesis, making it the least likely interpretation of the data. Though, if a sensible definition of biochemical damage for misense mutations can be produced (one has not been forthcoming), it is possible these mutations are damaging. So his hypothesis might survive with new data, but right now it is severely challenged by the polar bear data.
Where does that leave us? I’m not sure. None of this information is new. I’m would like to think that Behe and ENV can admit error, like all good scientists do. Failing that, this post should clarify for all students following this the scientific errors.