Thursday, May 31, 2007

Good Virus, Bad Creationist

Man Mark over at Good Math, Bad Math got an advanced copy of Behes latest, 'The Edge of Evolution.' (Foiled by D-list blogging again-- Deniers didnt send me an advanced copy of Blondies 'Science Outsold'. I even asked!)

Now, Behe has always aggravated me, as I first became aware of his existence when I was taking introductory immunology... and I found I could refute his 'arguments' regarding the immune systems. "That dude isnt right" I said to myself at the time.

Ive tried to ignore him as best as I could since then. Honestly, I thought it was a little funny that he went after the bacteriologists with the damn flagellum and the immunologists with "Immune system cant evolve blah blah", but I never thought he be silly enough to go after something I research. I mean Im sure he reads my blog daily and knows hed be snarkified by a chick half his age if he tried to touch stuff I do.


GM/BM-- Anyway, the new book is based on what comes down to a mathematical argument - a mathematical argument that I've specifically refuted on this blog numerous times. I'm not mentioning that because I expect Behe to read GM/BM and consider it as a serious source for his research...
*giggle*

Okay, so I cant claim 'D-list blogger help help Im being repressed' this time. I guess Behe doesnt read much of anything, if Marks summary of his latest book is accurate.

The basic argument in this chapter is the good old "fitness landscape" argument. And Behe makes the classic mistakes. His entire argument really comes down to the following points:

  1. Evolution can be modeled in terms of a static, unchanging fitness landscape.
  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.
  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.
Ouch. I dont talk about my research directly a whole lot here (except for pretty pictures, of course), but like I put in my blurb, I study the evolution of HIV within patients and within populations. Fitness and fitness landscapes are vital to my research. And if Mark has summarized Behes claims properly-- Im kinda peeved *fumes*

No one can have a basic, basic, basic understanding of 'fitness landscapes' and come out thinking those four points are valid. Just watch, Ill explain fitness landscapes to you all right now in the context of HIV, and you will get it! You, even those of you with zero biological training, will be able to refute Professional Creationist Michael Behe! Yay!

Okay, check it out-- think of a graph. On the Y-axis, think of something that contributes to natural selection. For HIV, a good one is T-Cell escape. On the X axis, every point is a potential HIV sequence. EVERY possible HIV sequence. X-axis is real long, but lets pretend it looks like this: The HIV genome sequences that are 'better' at staying under the radar of T-cells have higher Y-axis values (peaks), and the ones that cant escape very well have lower values (valley).

This is a two dimensional graph. Its not really two dimensions. You see, for every contribution to natural selection, you add another dimension. Add a Z-axis to that graph-- escape gp120 antibodies. Add a W-axis-- escape gp41 antibodies. Add how fast Env can get the virus into the infected cell, host nutrition, resistance to Drug 1, resistance to Drug 2, resistance to TRIM5a, resistance to APOBEC, etc etc etc etc etc. Physicists get excited when they find another dimension... but they aint got nothin on HIV fitness landscapes. I dont think you could even quantify how many dimensions there are to that simple 2D graph from above.

So Sequence 1 might be GREAT at escaping T-cells... But it might be awful at controlling TRIM5a. Sequence 2 might be incredibly efficient at infecting cells, but isnt resistant to Norvir. So HIV explores all these possibilities and creates a quasispecies-- instead of points on a graph, like above, when you combine all the possible fitness axes, you get a cloud of maximally fit viruses. NOT optimally fit! Maximally fit, for the given parameters!

This is what you find when you take a blood sample from an infected patient and sequence all the HIV viruses you find-- Most viruses have a very similar sequence, but different enough that you get a cloud, not a single point of identical viruses. Let me amuse you and draw a picture in paint:
But this picture is just a snapshot. One week later, the patient might try a new drug-- the quasispecies will shift. One week later, the patient might transmit a virus to her partner-- quasispecies is presented with an entirely new landscape.

So:
  1. Fitness landscapes are never static. At least I cant think of a scenario where that would happen. Ever. Someone give an example if you can think of one.
  2. A peak in one dimension can be a valley in another dimension. This is clearly demonstrated in HIV, where drug resistant viruses obviously have a fitness cost. When you take the patient off the drug, drug resistant HIV is rapidly overtaken by wild-type viruses.
  3. What is it with Creationists and things always getting 'better'. Right now Im trying to determine if fitness increases/decreases/randomly changes over time in long-term drug naiive HIV patients. For all I know, HIV viruses are decreasing in fitness over time, but increasing in transmissibility. Things dont always go in one direction!
  4. Dumb, dumb, dumb. One/Two/Three base-pair changes in HIV can tank a sequence from the top of a peak in one dimension (replicative capacity) to the bottom of that dimension, and pop you to the top of the quasispecies in another dimension (drug resistance). One mutation doesnt equal a little better, another mutation equals a little better, another mutation, etc etc etc.
There ya go folks. I promised you all could understand this, so if you dont, please leave your questions in the comments!

26 comments:

Tyler DiPietro said...

Just so you know, the invitation I extended to blog at GGA is still open. If you want to do topical stuff over there you could always do mathematical biology posts like this!

ERV said...

Hey speaking of-- You really should go into math-bio if youre concerned about job security.
The demand for folks who know math tricks and biology is going through the roof with modeling of pathways and proteins and populations and epidemics and all sorts of cool stuff. Its so big my university (small, liberal arts) started a combined math-bio program.
I mean, at my school you majored in 'Biology' or 'Chemistry' or 'English', so to have a specific program like this is pretty big.

Tyler DiPietro said...

It's certainly something to look into, thanks for the suggestion.

One thing that worries me about fields that go through booms like that is they tend to burn out pretty quickly. See the dot com boom in the IT industry or the more closely related biotech boom. The hype soared and the market got saturated as qualified people poured in, leading to big lay-offs.

But I will definitely consider it as an option, I can probably still safely take on another major. It certainly seems more attractive than physics rendering and geometric modeling, which is largely project based and thus not really a first resort for job security.

RBH said...

Apropos of this comments discussion, the bio department I"m associated with has a guy on joint appointment in math and bio, and it's in a private liberal arts college with 1500 students. My best guess is it won't burn out in a few years.

Dustin said...

I doubt that math-bio will ever go away. As molecular evolution heats up, we'll need better and better computational methods for dealing with the subject. Population genetics and dynamics are intrinsically mathematical. Physiology relies so heavily on fluid mechanics and physics that our department works on the topic routinely.

Plus, thanks to the abuse of antibiotics, I doubt that the future of epidemiology is ever going to be in question.

For my part, if it doesn't have words like "contravariant" in it, I'm just going to hamfist my way through. I stay as far away from computationally oriented stuff as possible. I'm not contemptuous of it, I'm just inept at it.

Dustin said...

Oh, and Adolphe Quetelet once said:

"The more progress physical sciences make, the more they tend to enter the domain of mathematics, which is a kind of centre to which they all converge. We may even judge the degree of perfection to which a science has arrived by the facility with which it may be submitted to calculation."

Biologists (and even physicists, believe it or not) usually flip out whenever I throw that quote out there -- but it's true. Einstein put it a little more humbly, suggesting that mathematics gave a measure of security which we could not otherwise obtain.

Still, math-bio is here to stay.

Torbjörn Larsson said...

Nice post, and anything that needs "quasi" in it is certainly intricate enough. ;-)

One nitpick though, physicists deals with dimensions of phase spaces too. In fact, I hear that quantum field theory need infinite-dimensional spaces to work.

Dustin, that quote is a good basis for a long discussion. (At a more suitable location, perhaps.)

For one thing, Polchinski (of string theory fame) points out that rigor is somewhat at odds with easy progress, and that links in theories and between theories give sufficient support without it.

But certainly, the more basic possibility for quantitative descriptions is instead an important tool for progress and increased security. So math is certainly good but beware of pushing it to rigor mortis.

VancouverBrit said...

ERV, you work on some really cool stuff! Anything published yet? I'd love to read it in more detail!

ERV said...

Torbjorn-- Oh fine, squash my fun poking at the physicists :P But thanks for the correction!

Vancouver-- Nope. Its just me and Bossman in the lab, so Im doing all the plug-and-chugging myself. Hopefully Ill have a paper very soon, though!
The best Ive got for you now is a link to a technical paper that we sort of let our imagination run away with. We took this guys system and made it better, and put it to some good use **big grin**

Use of a novel assay based on intact recombinant viruses expressing green (EGFP) or red (DsRed2) fluorescent proteins to examine the contribution of pol and env genes to overall HIV-1 replicative fitness.

Dustin said...

Dembski is throwing a tantrum. No comment on how or why Mark is wrong -- just the assertion that he is wrong. But, then, proofs aren't Dembski's cup of tea.

ERV said...

HAHAHAHAHAHA!

A bottle of single-malt-scotch says 'Sandford' is Dembski! AAAAAAHAHAHAHAHAHAHA! THATS SO HIM!! HAAAAAHAHAHAHA!

Where is Dembski holed up these days?

Dustin said...

"Where is Dembski holed up these days?"

Ann Coulter's Hitler-fetish dungeon.

ERV said...

There shall also be no condemnation of fetish dungeons on my blog. :P

Dustin said...

I'm still waiting for my relatives to figure out why I spend so much time at the hardware and upholstery stores, yet they never get to see any of my projects.

Anyway, I shouldn't have said that. Coulter and Dembski probably don't even copulate (let alone doing it in a fun way), favoring sporulation instead. Pod people.

Nick (Matzke) said...

This post and all the posts on this blog are friggin' hilarious. Just linked it from PT. Keep it up.

Bob O'H said...

Right now Im trying to determine if fitness increases/decreases/randomly changes over time in long-term drug naiive HIV patients. For all I know, HIV viruses are decreasing in fitness over time, but increasing in transmissibility.
Well, natural selection means that fitness will increase in time (Fisher's Fundamental Theorem and all that). If the virus is decreasing in fitness, it will be selected out.

I guess you're only looking at some components of fitness (virulence? growth rate?), and there can obviously be trade-offs, and selection acting at different levels.

Bob

Anonymous said...

Cool article.

Sweet MS Paint skills sister.

typo note: tank = take?

ERV said...

Bob-- Sorry, I should have defined that. We use the term 'fitness' exclusively for reproductive capacity. With HIV, reproductive capacity is directly correlated with how fast a virus can get into a cell, a process mediated by the gene 'env'.

Of course we have to asses viral fitness in an artificial 'vacuum' of sorts to avoid the problems I pointed out in this post-- there are so many variables effecting fitness that we couldnt possibly tease out the dimension we're interested in in (env) an in vivo system. T-cell escape, antibody response, etc. Its hard to get down to a 2D graph.

It could be that 'env' is getting 'better' at infecting the cells of the host as time goes by, but is losing some ability to cross mucosa membranes (transmission). Or maybe transmissibility is increasing too. Or maybe transmission is completely random :) Wont know till we look!

You know, I think Ill just do another post on HIV quasispecies and fitness-- I cant get in everything I want to say to you in a comment, so good question :)

ERV said...

Nick-- Thanks! I really appreciate it, as I use this blog as a way for me to learn new things and master older things Ive learned and forgotten and such. This post is going to make me go back and read 'Quasispecies: Concepts and Implications for Virology' to answer questions... something I would not have done voluntarily before... hehe!

Anon-- The only thing better than my Paint skillz is my proof reading skillz :P

Torbjörn Larsson said...

"Oh fine, squash my fun poking"

Aww... Sorry about that!

Btw, I rather imagine squishing it. I think it's the funny sound I prefer.

"typo note: tank = take?"

Probably not; it works either way and is IMHO more consistent with "pop up".

"tank" := "Slang To suffer a sudden decline or failure" ( http://www.thefreedictionary.com ).
"pop up" := "appear suddenly or unexpectedly (ibid).

ERV has a colorful imagery.

Torbjörn Larsson said...

Oh, seems we crossposted.

Perhaps I read into this to give ERV more skillz than she admits to.

Either way, I liked the alternate reading. :-)

ERV said...

hehe Nononono I meant tank :) But click back on some of my older posts-- pre-Firefox-automatic-spell-check.

Its good for some laughs :)

Theres a joke in my hometown-- all the kids that went to the Catholic elementary school can spell, but they cant add. And all the kids that went to public school are great at math, but have trouble spelling 'Wednesday'.

Its so true.

Duncan Buell said...

You write, "See the dot com boom in the IT industry or the more closely related biotech boom. The hype soared and the market got saturated as qualified people poured in, leading to big lay-offs."

Although it's somewhat off topic, it's a calumny that should be rebutted. In fact, the job market for computing has been better through this decade than the job market overall and it has been better than the job market for anything else in science or engineering.

The problem has been the spin and misrepresentation by the media, which has led people to think that in fact the market has gone sour. Computing has a projected 1.4 million jobs over ten years, with fewer than 550,000 bachelor's degrees to be produced. This is contrasted with all the rest of science and engineering (health care excluded), with 600,000 jobs projected for 1.6 million new degree holders.

Two anecdotes that illustrate the problem. About 2001 I saw a headline that forecast a decline by 44% in the number of vacancies in computing. This was true; the projected number of vacancies went from 1.4 million to 950 thousand or so. What the headline failed to mention was that the number of jobs filled went from 250 thousand the previous year to 400 thousand the current year. So what had really happened is that the number of jobs that went unfilled dropped by a huge amount.

Second anecdote: A year ago the Chronicle had a story about declining enrollments in computing. They interviewed a graduating senior who said he had received three job offers in his last three months as an undergraduate. And he was deeply disappointed in this. Perception can be everything.

Shawn Wilkinson said...

I learned something new a valuable today. Thank you, oh wise one ;-)

BTW< as a math-bio person, do you think Evolutionary Dynamics by Nowak is a good book? I'm a Chemical Physics person myself but want to expand my horizons, as they say, so I'm looking for the opinions of those who actually work in the field for good starting texts.

Also, as a future science educator, I find it disgusting for pseudoscience gaining in popularity.

/ranting

ERV said...

Oh gawd Shawn! I havent taken a math course in years and years! Im not a math-bio person :) Thats the thing thats confusing me so much about this whole ordeal.

Yes, there are mathematical models for fitness landscapes. But what I do in the lab is not theoretical. Im taking 'Polaroids' of fitness landscapes and population frequencies at various points in time and actually watching how everything is moving around. Its not a model, its reality.

So I have no idea what the good math books are, but I can hunt you down some good 'biology' fitness landscape papers!

Shawn Wilkinson said...

I see. Don't worry, I won't do what they did in South Park and killed the Wise One for not giving the desired answer ;-)

But all and all, much thanks for giving good reads on fitness landscapes and their practical utility.