Okay, Ive gotta admit, Ive been royally confused at UDs response to Mark Chu-Carrolls review of Behes new book. That they would attack his 'credibility' and pretend thats a rebuttal of his arguments isnt confusing. Whats confusing is that theyre unimpressed with his mathematics credentials. Not his biology credentials. Math. And theyre desperately attempting to give Behe some ethos as a mathematician (eg '... there was more math in chemistry than in computer science.')
This is very, very odd to me.
My boss is a biochemist (BS and PhD). Everything I do is related to biochemistry even though I dont consider myself a biochemist. My boss and I use fitness landscapes.
Behe is a biochemist too. So, whats his excuse for screwing this up so badly? From my point of view this has nothing to do with 'math' at all.
If anything, I thought UD would go after Marks lack of biology credentials. I know that Dawkins WEASLE program (see Dysfunctional Analysis and Greedy Greedy Algorithms) and such use genetic algorithms and all sorts of fun math that are well beyond my Calc I capabilities... but if 'easier' biochemical (non-math) examples refute Behes statements, whats the point in damning someones mathematical credentials?
As I pointed out in a comment on PandasThumb, nothing I do regarding fitness landscapes requires high-level math. And its not just me-- let me go over a recent paper in Journal of Virology to give you all another example of how Behe is being extraordinarily silly (or maybe this is his normal level of silliness, meh.) Nothing in this paper goes beyond simple algebra and statistics (except for the math behind phylogenetic tree programs).
Fitness Landscape of Human Immunodeficiency Virus Type 1 Protease Quasispecies
I know all of you can read the abstract, even if you cant see the whole paper, so you know where Im going to go with this... but lets not play Creationist and say a paper supports our views just from what the abstract appears to say.
Okay, an essential protein to HIV is called 'protease'. Protease is an enzyme responsible for maturing new HIV viruses after they bud off of the host cell. If there wasnt a maturation step like this, there would be nothing preventing new viruses from turning around and infecting the host cell again (all the right receptors are right there). So this makes the protease enzyme a great target for anti-HIV drugs. Youve probably heard of 'protease inhibitors' before.
Unfortunately because of the quasispecies nature of HIV, the second you give someone protease inhibitors, HIV finds a way around that hurdle, so PIs cant knock out HIV for good. What makes this even worse is protease can mutate a LOT and still function just fine... or better than the quasispecies consensus sequence. What Creationists say cannot happen: "... rugged HIV-1 protease quasispecies fitness landscape may be formed by a continuous network that can be traversed by single mutational steps without passing through defective or less-adapted proteins."
So without any slights of hand, this is what these researchers are going to do, no intensive math required:
"We determined the fitness of each variant present in the quasispecies in order to establish the relatinships between genotype, phenotype, and fitness and constructed a phylogenetic-fitness landscape map for each quasispecies."Step 1: Filter out all the dimensions you arent interested in.
First thing these researchers had to do was get the protease gene out of its natural environment. Take it away from the CD4+ and CD8+ T-cells. Away from the antibodies. Away from in vivo. An easy way to do this with retroviruses is to blow up some infected cells, isolate the cellular DNA, put one copy of cellular DNA per well, and PCR out the gene youre interested in. Each infected cell will yield a different protease gene. Now, the sequence you isolate might be a dud. It was 'good enough' to get in, but the reverse transcription process mutated it out of working order. Thats okay, in this case. We want to see the duds (non-functional proteases) too, to determine why they are non-functional.
Now again, we have to strip all the other dimensions of the fitness landscape away, so you put the isolated protease genes in a common background- in this case a bacteriophage.
Step 2: Test the dimension youre interested in.
Infect cells with phages containing different kinds of proteases (note: theyre using E. coli here to use phages, not eukaryotic cells!). The special bactereophages allow the researchers to compare protease activities-- its a model system instead of directly measuring protease activity differences (they go back and make sure their model predicts normal HIV protease activity. it does).
Step 3: Combine your fitness measurements with your phylogenetic tree.
Ill take this down if JViorl wants, but I think its fair use:Bigger the circle, higher sequence frequency. Longer the line, more different the circle is from other circles. The different colors relate to how active each protease sequence is. Translation: You can be real different from the consensus sequence and still work fine.
Breakdown of what percentage of sequences they found that were 1, 2, 3, 4, 5, 6 base pairs different from the consensus sequence!
Diagrams of protease color coded to show where mutations were occuring, and at what frequency!
Relation of fitness in the bactereophage system back to fitness in HIV!
Relation of replication capacity to protease activity in HIV!
Super cool relation of mutation rates and protease activities back to the disease progression in the patients!
No hard math!
Oh wait, Im supposed to write about why Behe is silly, not why this particular paper is cool.
Okay, let me go back to the four points, and put some relevant text from this paper for a rebuttal. *sigh*
1. Evolution can be modeled in terms of a static, unchanging fitness landscape.
"... although the three analyzed quasispecies shared some traits, such as the presence of several fitness optima, every protease quasispecies formed distinctive individual fitness landscapes...
... even quasispecies N and O (two different patients they looked at, ERV), which had similar amino acid diversities, formed very different fitness landscapes."
2. The fitness landscape is a smooth, surface made up of hills and valleys, where a local minimum or maximum in any dimension is a local minimum or maximum in all dimensions.
"... within each quasispecies, there was a large number of fitness optima... Because fitness optima were frequent, the HIV-1 protease quasispecies complexity does not depend exclusively on the enzymes catalytic efficiency..."
3. The fitness function mapping from a genome to a point of the fitness landscape is monotonically increasing.
4. The fitness function is smoothly continuous, with infinitessimally small changes (single-point base chanages) mapping to infinitessimally small changes in position on the fitness landscape.
...Several single substitutions were lethal and led to the master sequence to drop down the peak. However, at other positions, single substitutions sent the master sequence to a new optimum or peak, suggesting that the master sequence may walk through the quasispecies fitness landscape without being trapped at suboptimal alleles...
No advanced super-hard math tricks that no one can understand except for 'math' PhDs.
What am I missing? Why are the UD-crew going after Mark Chu-Carrolls 'math' credentials?
Edited 8.30 pm 6/4/07-- I realize Behe tries to use math to support his views in the chapter Mark is discussing, but Behe should know, as a biochemist, that his arguments are incorrect. No math should be needed... unless hes trying to intimidate his readers into thinking 'Oooh hard math! He must be right!'