Monday, February 1, 2010

Les Assises, late Saturday

Well, actually I'm behind in my blogging and the meeting is completely over. The poster is back in its tube. My USB stick is gone, forgotten in the meeting's computer after a quick download of rabbit pictures before the morning session. And my camera's memory card didn't record a single one of the photos I took on my walk around Saturday afternoon in the last of the light. You'll never see the sunset over the Rhine. Er, not my photo of the sunset over the Rhine. Possibly I just didn't put the card all the way in its slot or something. Possibly there will be nice pictures of Strasbourg Sunday morning, but it just depends. My train is at half past 11, so if the card behaves there's hope.
All that is not the point of this blogpost. It's just the usual messing around convincing you of the painfulness of my life when it's really nothing much. My hotel might be cheap, but it's clean and the mattress is comfortable and I have no loud neighbors.
Back to a bit of science.
I mentioned on Friday's post some of what you can do now that you can look at hundreds of thousands of markers in tens of thousands of people. This sort of massive genetic search is being used to work out the genetic contribution of very complex diseases, such as Type 2 diabetes and heart disease and hypertension, which have multiple environmental factors and yet a clear genetic influence.
Type 2 diabetes genes are big treasure these days, and many huge international consortia are searching them out. One of these studies was presented to us on Friday, where they took a good many people (I didn't write down all the details, but they're in the paper) with T2D or with a particular insulin & glucose level response 2 hours after eating a meal. They looked at markers over the entire genome to find those associated with an abnormal response. And they found some.
Frankly, if you look at many thousands of markers, you're always going to find something more or less correlated with the anomaly you're testing. If the odds of getting a result at a marker are 10,000 to 1 against (absolutely Fabulous in the world of genes!), for every ten thousand markers it's an even bet you run across a hot one just by chance. So we take the markers that pass the statistical tests with a grain of salt.
(I just realised that I've eaten in this restaurant before. I was in Strasbourg 3 or 4 years ago, and we landed in a tiny little place at random, the very day before they were closing up shop to leave on vacation and they were emptying the fridges. What a meal! It was wonderful then and it is now. I was actually thinking about it yesterday, hoping to run across something similar, and here I am - at the restaurant literally next door to my hotel! What threw me off is last time we entered through the door at the other end of the long narrow room, off the parallel street. Just 8 tables. They can seat 25 and that's it. It's Jean-Paul! Every table has a pair of salt-shakers that are unique, like you'd find at the flea market. My table has chipped chickens. Across the way it's cats, and next door it's cows in chef's hats.)
(The mousse au chocolat that I've just ordered is known as the Small Dog. A sort of Doberman cross. Mmm. I don't really get it either but apparently it's a treat.)
Weren't we talking about genes for diabetes?
Ah yes.
Well, some of the candidates involved in glucose metabolism have already been fingered, but they account together for maybe 10 % of the genetic contribution. And those are the best candidates, so you see that now we're looking for little bits.
So fine. The genome-wide screen pointed up a new handful of loci possibly involved.
That means that a marker came up hot; but then they decided, kind of just like that, that whatever gene was closest to the marker must be the important one.
It doesn't always work out that way. Especially now that we know that DNA sequences that aren't genes in the traditional sense can have enormous effects. But the top marker was somehow related to glucose metabolism, so it's easy to be seduced by that connection. They may well be right, but that kind of jump can take you off on a wild-goose chase, and when you eventually throw out the result as spurious, you're so sick of it that you're not likely to go back and look for a different explanation for the initial result.
(As yes. The dog connection for the chocolate mousse. It's served on a plate... Three large spoonfuls... sort of spread out....It's just as well he makes the joke in advance.)(It is really good, though.)
What really bugged me about the study, though, was their strategy for continuing. They decided that a gene really involved in diabetes must increase in RNA expression in response to a meal.
Well that's just silly. It's well known that the level of RNA made for a given gene can have relatively little correlation with the level of protein that is made, and still less with how active that protein is. (DNA codes for genes, which are copied into an RNA intermediate, and the RNA is translated into the protein, which does the actual work the gene is known for. But each RNA molecule can be read many times, and the protein itself can be more or less stable depending on conditions. Proteins also move in or out of cells, or change their compartments, or acquire a phosphate group, or etc etc in response to a stimulus.) So the researchers had this a priori idea that a role in diabetes had to necessarily pass through an increase in RNA in order to have anything to do with it. And when the RNA level didn't budge two hours after a meal, they decided that meant it was just another red herring.
I wasn't enamored of your skip in the chain of evidence above, but if it really is a logical candidate, there's no need to be so hasty about tossing it in the dustbin!
The point here is: just be careful about what you're doing. Mind those fixed ideas about how your new gene "should" behave. You're there to discover how it behaves, not to document your preconceived ideas. Mind the chain of evidence as well. If a clue leads you to the middle of a street, you can't just pick the prettiest house as the right address, even if there is a sign in the window saying Please Come In.
Um. Strike the loud neighbors thing. Just for tonight, there is somebody playing the piano on the other side of the wall. Not part of the hotel, so it’s harder to yell at them. It’s decent playing, but it just goes on and on.
Pictures tomorrow.
Though if you just can’t wait, there are already a few over at Pink Rabbit Abroad.


Bagman and Butler said...

I understand your frustration about some of the research being done on red herrings these days. But it is so hard to keep the research grants going that to keep getting paid, some scientists are forced into coincidental minutae. I remember being at a huge NIDA conference where some of the best brains around were locked in a discussion of the scales on a chart on a Powerpoint slide. I finally stood up and asked, "With all the brainpower in the room, shouldn't we be looking for an actual cure for alcoholism?" Given that I have no letters after my name to justify such a smartbutt mouth, I got nothing but blank stares. But in the hallway afterwards several people came up to me and said they thought I was right on but were afraid to say it.

NanU said...

One of the hardest things to teach my students about science is that you don't do an experiment 'to prove that...'; you do it 'to see if...'.
Huge difference!