Tuesday, January 24, 2017
Every so often you'll read something about how genome sequencing has gotten cheaper, or how genetic engineering has led to a breakthrough in certain agricultural products. But then you'll think to yourself: "The Human Genome Project" was done at least a decade ago. How come we haven't gotten anything interesting out of it? Has there really been zero progress? Where are the gene-engineered babies and super-athletes that's going to make doping obsolete?
The Gene: An Intimate History goes a long way towards explaining what's going on in the field. It's biggest problem is that it's a layman's book. So in addition to having to suffer through the personal stories of the author and his family (yes, it's relevant, and if you're an English major you might find it interesting as it adds personal color, but I just rolled my eyes and skimmed it as quickly as I could), you have to suffer through the pedantic explanation of the discovery of genes through Darwin, Mendel, and various other folks like Galton. There's also significant coverage of the eugenics movement and the horrors of world war 2. If you're a reasonably well-read engineer this is all old hat and you can zip through as quickly as you can read.
The story gets interesting only when you get to Watson, Crick, Wilkins and Franklin. From then on, the exposition gets far more interesting as we start to explore the current knowledge about DNA, it's relationship to RNA, the relationship to epigenetic markers (which I didn't realize were real markers with chemical traces) and why progress in gene-engineering has been so slow.
Part of the problem is that the number of genes in the human genome is surprisingly small (19,000-20,000). You might think that this is good news, as that means that there are fewer genes to study and make sense of. That's not correct. Genes in the DNA are activated and de-activated as needed, and used to generate proteins. The problem with a much smaller than expected set of genes is that it means that the genes are probably used in multiple places in surprising ways with complex interaction between them. In other words, you'd much rather read 100,000 lines of well-structured code than 20,000 lines of spaghetti code, some of which modifies itself (or is used to generate code that then generates more code!). For instance, the human immune system has to have generic approaches to creating and reacting to anti-bodies, since at the start it cannot know which types of viruses or bacterial infection it has to respond to.
The other problem is that gene expression is not 100% all the time. The biological term for this is "penetrance." In other words, if a gene's presence causes a disease only 50% of the time, it's not enough to detect for the presence or absence of the gene. You also have to understand the environmental triggers that cause the disease in the presence of the gene. That's a problem even for single-gene diseases where a distinct gene can be tracked down that causes say, Huntington's disease, or certain forms of breast cancer. It's an even bigger problem for multi-variate factors like intelligence, where multiple genes from all over the genome might contribute. In other words, unlike genes for hair color and eye color, most genetic determinants of attributes we care about cannot be tracked down to a single gene, which makes everything much harder to develop.
Then there's the editing problem. Until relatively recently, there's been no easy ways to edit a gene sequence. So even if you did know the changes you want to make to a genome, you'd have no way to edit precisely the change you wanted to make. The barrier to this is slowly falling as new techniques are developed, but even with the new techniques the delivery mechanism to an adult human is full of danger: previous gene therapies have been tried which have killed the patient.
Finally, there's no complete model of the human genome as it interacts with the environment. This is a severe problem, so editing a gene could have unintended side effects and consequences. It boggles my mind that there isn't a project to provide a computer model of the human genome from the DNA up. You would want there to be a "virtual human" the way there's a "virtual machine" that lets you experiment when you build a new operating system or to see the changes you make. (Or at least, maybe there is such a project but the author of the book didn't see fit to mention it) Until that kind of technology is available to at least predict what your changes are going to end up doing, gene engineering seems kinda dangerous, like writing code on a machine with no process isolation --- any mistake could end up killing the patient!
All in all, this was a decently comprehensive book, and does a great job of explaining why we don't have super-intelligent engineered babies or super-athletes that don't need doping to win. Recommended.