Hacker News Reader

Comment by SteveGregory | Hacker News Reader

Gene drive is a very scary thing from a strict biological standpoint. Just imagine a gene that must be passed on to all offspring, rather than randomly chosen. We could quite easily wipe out entire biomes with such a technology.

This should not be underestimated, or thought to be less relevant of a weapon than, say, the atomic bomb. It's just slightly more subtle than explosives.

Yes, there may be benefits if used well. And yes, nuclear technology can be (very) useful for generating power. I'm just saying that there needs to be more attention here. The particular uses need to be more widely understood and talked about.

View on HN

It's a common breeding technique to create individuals with identical gene pairs so that all offspring will be guaranteed to have the gene. This is what "purebred" means.

https://en.wikipedia.org/wiki/True-breeding_organism

The difference here is that even if bred with a non carrier, all offspring will be homozygous, and all of their offspring ad nauseum. This is very different than a general that can be diluted or selected against.

Not the same at all.

An individual homozygous at a given loci has no guarantee that its offspring will also be homozygous. It will have one copy from that parent, but the other parent may or may not provide an identical copy. If it's a new trait (say a doomsday gene), then no other copies exist in the breeding population, and all offspring will be heterozygous. If the overall population is static, then average expected fitness is two offspring that make it to breeding. Thus, there will be two copies of the 'doomsday' allele in the population, just as there were in the original homozygous parent. In other words, the total number of alleles remains static over time.

In order for an allele to take over a population (become fixed), it must confer a fitness advantage, or else make it there by pure chance. Relying on chance, you have to introduce huge amounts into the population so it becomes the majority, otherwise it will be eliminated.

Gene drive is completely different as these constructs can copy themselves to the homologous chromosome in heterozygous individuals. Offspring will be homozygous even when only one parent contributes the allele, and thus the expected allele content in the population will double each generation. Now only a few individuals need to be introduced into a population for the trait to become reliably fixed.

It's exponential, and it's truly terrifying.

If the gene makes individuals less viable, the distant cousin without the gene is going to outcompete carriers of the gene. Unless we come up with a way to inject a gene into all individuals in a biome simultaneusly, but in that case all offspring are going to have the gene, drive or no drive. If the gene makes offspring more viable, how will it wipe out anything, except maybe the individuals without the gene?

That's not how evolution works. Evolution does not select individuals, it selects genes.

Let's say that one of these genes kills offspring with probability 25%, but has gene drive so it's present in 100% of offspring. A normal gene doesn't kill offspring, but is only present in 50% of the offspring. The introduced gene will spread rapidly, since it has a 100% x 75% = 75% chance of being in a living offspring, even though it kills, since the ordinary gene only has a 50% x 100% = 50% chance of being in a living offspring.

That's why this is so terrifying. You can introduce a gene into a population that kills off the whole population.

https://en.wikipedia.org/wiki/Gene_drive

Actually is the gene is too efficient at killing its offspring, it will kill its bearers before it can spread, and if it's not lethal enough, the population will just deal with it.

At 25%, it should allow the gene to spread in the population, but whether it's enough to eradicate the population depends on a lot of thing (the environment, how old the individuals die, etc.)

natural factors already causes a fairly high mortality of offspring before they get to reproductive age, and it seems like very few species would be able to cope with an additional 3-of-4 culling for an indefinite period of time.

You're missing a very critical detail.

Normally, for a trait to grow in a population, you need a fitness advantage. Basically, if you average a little over 2 offspring that successfully breed, that trait should grow to fixation.

In the case of exactly 2 reproducing offspring, allele count is static. Homozygous parents will provide one allele to each offspring, so 2 copies in one generation, and 2 copies in the next. Heterzygotes have a 50% chance, so one offspring is expected to have the allele. 1 copy in the parent generation, 1 copy in the next.

Gene drive changes the completely, as the construct will copy itself to the homologous chromosome. You'll go from 2 to 4 or 1 to 2 copies from the previous example. The growth is exponential. There is nothing like that in nature.

Thus, even if a moderate fitness disadvantage is introduced, the allele can still be driven to fixation quite easily. Even with an expected offspring of 1.5, you can still go from 1 to 1.5 copies of the allele over each generation.

The other issue is that you can introduce a trait for e.g. chemical susceptibility. It offers no fitness disadvantage until that chemical is introduced into the environment.

Reply on news.ycombinator.com