Scientific American had a very good article on the genomics revolution, ten years after the first human genome was fully sequenced. That sequence was a signal event in the progress of human knowledge, as was the prefiguring discovery of DNA's structure fifty years before. But who has benefitted? There is no comfortable way to say this ... biologists have made out like bandits so far. They (especially Francis Collins) promised a transformation of medicine resulting from the knowledge gained through this sequence, but that train has hardly pulled out of the station ... yet.
Biologists have seen their field transformed by this new knowledge and associated technologies, which have multiplied their power to ask questions of organisms. Where they were happy to know how one or two genes responded to a drug or other condition, now they measure the response of every gene in an organism. Where they spent years mapping the mutation and gene responsible for an interesting phenotype or disease, now it may take only days to figure out the gene's location.
What's the slowdown? Obviously, it has to do with complexity. Even with the genome in hand and with accelerating technical capacities, understanding doesn't come as rapidly. The blueprint metaphor for the DNA code doesn't quite work because it is not actually a scale model or drawing of the organism's morphology, or a neat drawing of its chemistry either. Instead, the DNA is a digital code for rune-like protein sequences that assemble themselves into dynamic animated super-heroes that fly around the cell doing difficult-to-understand jobs, controlled not by a single mastermind, evil or not, but by a bogglingly complex network of whatever happened to crop up during evolution. Perhaps a sprinkle of protein phosphorylation here, a location-specific DNA controller there, and indirectly regulated protein degradation as well. It all goes into the pot of what makes our biology go.
Integrating all this action and re-action conceptually, as the cell does in the flesh, has been an enormous problem that has made computer sciences particularly important in biology. But success in that kind of modelling remains tantalizingly far off. Biology is in essence an alien technology, perhaps the most alien technology we will ever encounter, having nothing to do with our macroscopic technologies of carpentry, stone-hewing, metal-working, etc. Its principles are completely different, and not simple.
Another issue is about what is even possible with respect to medicine. Not every malady will ever be addressable by a drug in the current paradigm, which is making drugs that interfere or help the function of proteins. Only when we gain the capability of using drugs to directly control and change genes throughout the body's DNA will our pharmacological powers truly be omnipotent, something still quite far off, and of course full of ethical conundrums as well.
But back down to earth... the most interesting point made in the article was about a push over the last decade to map common variants in the human genome, among people and populations, in hopes of identifying the most important disease-causing genes and pathways. (One example is the HapMap project.) Because sequencing one genome was of course not enough- we had to sequence lots of genomes and compare them to figure out how people differ, in all those interesting ways that make us happy or unhappy. Eventually, every person will have their genome sequenced, which will be a rich and central part of the individual and collective medical record.
But the logic of this first variant hunt was fatally flawed, for eminently Darwinian reasons. The idea behind it was that many traits, like hypertension or diabetes, are common. Thus the genetic variants that cause them should be common as well, making them easy to detect with the modest sequencing technology of the day. But actually, this was like looking under the street light just because you can, rather than because the keys are there. Genetic variants become common in a population precisely because they are inoffensive, perhaps even fitness-enhancing, not because they are the primary causes of the diseases that doctors and geneticists were hunting around for.
What the researchers found, at the cost of about $100 million, were lots of variants, but precious few with any effect on disease, and those with only tiny effects. They were essentially useless either for diagnosis or for the important work of studying genetic networks/pathways of disease. This led to a lot of papers (and news reports) about gene X linked to, say, alzheimers, which later turned out to be insignificant. It turns out that rare variants are the ones that are far more influential for diseases with genetic components. We all have lots of variants, and the ones that kill us are likely to be quite rare in the population, understandably enough. But more always come along, to be sifted out again and again via selection.
A similar story has happened with the specific condition of autism, which has been found to be related to variants in a stunning number of genes- dozens, if not hundreds. It looks like autism arises from defects in a broad developmental process in the brain that can be derailed into some characteristic groove by many genetic defects, each of which are naturally weeded out during evolution and thus never become common in the population.
So, sadly, ironically, but perhaps not surprisingly, molecular biologists had a weak understanding of evolutionary biology, were smitten by their new sequencing toys, and thought that the most accessible data (common variations) would be informative, or at least were able to convince their peers/funding agencies to that effect. The common variations have certainly been useful in some ways, however, such as tracing human ancestry over time and geography. And, now that we know the forces involved, in setting a baseline of sorts for low-impact genetic variation.
Now that rare variations are far more accessible as sequencing technology advances, researchers have been increasingly successful in finding genetic links to diseases. But what of it? One dream in the field was that genetic testing would be able to help us predict disease propensity with great accuracy. But if the disease-causing genetic variants are rare if not novel, and, in scientific terms, uncharacterized, they can't be used for a reliable prognostic test. This dream remains essentially unrealized. On the other hand, finding genes with stronger ties to diseases, by directly sequencing affected people and their families, helps the research enterprise tremendously, finally filling in some of the nodes in the networks that break down or go haywire in disease.
One example is a recent paper in Science by researchers who sequenced all the protein-coding genes of tumor cells from eight different human ovarian tumors. Amongst the scattershot mutations common in cancer cells, (these were not inherited germline mutations, but somatic mutations that arose during tumor development), they found four genes that had been hit more frequently. These four where then resequenced in a panel of 42 ovarian cancer tumors, 24 of which showed mutations in one gene whose product regulates chromatin accessibility. This gene had never before been thought of as a tumor suppressing gene, but its super-high rate of co-involvement here showed it to be one. Thus one more piece is added to our knowledge of how one kind of cancer can escape normal cellular controls. But knowing that this gene's inactivation can contribute strongly to cancer brings us little closer to a medical treatment, unless we have a way to replace its function ... i.e., to do gene therapy.
- Unlike basic science research, medical studies are a minefield of error as well as waste.
- You know it's hot when Ms magazine complains that not enough new atheists are women.
- Black and godless.
- Obama's unrequited good work.
- BofA and its MMT accusers discuss its fraud. Quite illuminating.
- Bill Mitchell quote of the week, quoting Joseph Stiglitz:
"JOSEPH STIGLITZ: My view is we cannot afford not to stimulate the economy. So, you know, anybody that says we should go back to austerity or we should not have a second-round stimulus just doesn’t understand economics. And let me be very clear about this. If we don’t stimulate the economy, the economy is going to get weaker. When the economy gets weaker, tax revenues go down and expenditures go up. Already, more than 40 million Americans are on food stamps. Number of people on Medicaid is reaching record levels. So, revenues go down, expenditures go up, deficits get worse. If you stimulate the economy, then people get jobs, they spend money, tax revenues go up. Now, if we spend the money on investments—investments in education, technology, infrastructure—you grow the economy in the short run from the stimulus, you grow the economy in the long term because of the returns that you get on these investments."
Bonus quote of the week, from Rand Paul:
"There are no rich. There are no middle class. There are no poor. We all are interconnected in the economy. You remember a few years ago, when they tried to tax the yachts, that didn’t work. You know who lost their jobs? The people making the boats, the guys making 50,000 and 60,000 dollars a year lost their jobs. We all either work for rich people or we sell stuff to rich people."
