Evolution Machine Accelerates Genetic Engineering 161
chrb writes "New Scientist has an article about the Evolution Machine — a device which can accelerate directed artificial evolution to discover desirable DNA changes in days rather than years. One of the aims of these researchers is to create an organism that is genetically immune to all viruses."
"genetically immune to all viruses" (Score:1, Insightful)
Someone needs to introduce these researchers to Gödel's incompleteness theorems.
We can't be trusted with this (Score:4, Insightful)
I wonder how easily it could be used to engineer the opposite case: a virus against which humans have no effective defenses.
Heck, just make on that takes out chickens, cows, and pigs, and humans all of a sudden have a major protein deficiency until alternatives (nuts, fish, etc.) could be ramped up, which would probably take at least 1-2 years.
Not without cost (Score:4, Insightful)
On the other hand, from studying influenza I can say that viruses evolve much faster than we do and if a variant (maybe adapted to another host) or subtype emerges that can bind your receptor anyway, then in effect you've selected out variants but not stopped the virus. Getting regular vaccines are still the way to go on this, IMO.
Welcome to the 1980s: (Score:5, Insightful)
This article is hyped up to the stars.
It's good work, but the ideas aren't "revolutionary" the way they are portrayed.
Lateral gene transfer in bacteria has been known for a long time. It's how resistance to antibiotics is spread among bacteria for example.
It's also been used a good deal already by microbiologists/biochem types (that line is getting a little blurred these days).
Church's group has found a way to automate this.
They can create large numbers of bacterial strains which have some or all of the desired characteristics in a short time.
The downside is the needle of the desired organism is in a haystack of partially successful or unsuccessful ones. In this case, it was linked to production of a bright red chemical. You could determine which was closer to the right one by color. That's easy to automate.
Most characteristics won't be that easy to screen or automate.
Church then goes into what's really an old idea. Encrypting the genome so that it's resistant to existing virus types. You then use a modified ribosome to translate that into proteins. I remember discussions of that in the late 80/early 90s on some of the transhumanist newsgroups (anyone remember usenet?).
The devil in the details here is that much of the information in the genome isn't for coding proteins directly, but for regulating gene expressions and other purposes. Much of that latter we still don't understand. It's hard to design an encryption to preserve a functionality you don't understand.
So, instead of throwing up their hands, Church et al appeal to using the above automated method and the microbes to sort out something that works, but again we really won't understand. At least at first.
It's an interesting idea. Sounds like a lot of work even if automated.
But, as anyone who was caught up in the genetic algorithms craze in computers can attest, it's not a guaranteed solution.
Re:but wasnt that just based on eugenics? (Score:4, Insightful)
What I'm saying with the above (and I hit submit before I typed this, alas) is that AA individuals are disproportionately impoverished or low income/education in the US, and consequently will be overrepresented in statistics that are influenced heavily by economic factors.
Look at any public health issue and it's actually income and education that play the largest role in disparities, not race. To try and paint it solely as a racial picture without acknowledging the very real role economics plays avoids the real issues at play and also leads to "solutions" that don't actually address the underlying causes.
Re:Welcome to the 1980s: (Score:4, Insightful)
Actually this research is really interesting...maybe not revolutionary, but interesting. I went to a talk by one of Church's postdocs at a conference recently and he was talking about this project. There are a lot of potential applications, but the example he was using was the optimization of the production of a metabolite. Traditionally this has been the hold up for synthetic biology. Getting microorganisms to produce industrially useful metabolites is not new. But engineering them to produce a large amount in and economical manner is where all the time and money goes because it requires some modeling, a lot of guessing, and mostly manual genetic manipulation. This technique uses the principal of directed evolution of a single gene (known for a few decades as you say) and applies it to an entire gene cluster, and potentially an entire organism. And it works! It's not a finished project, to be sure, but it can potentially become a very useful tool.
The "encrypting the genome" case refers to changing the codon code for the organism. Non-coding sequences won't be affected by it. The idea is that if you use a non-canonical genetic code for protein expression, foreign dna can still be inserted into the genome, but it can't be expressed. So viruses won't be able to replicate in the organism. It is immunity of sorts, but perhaps not really the way we normally think about it. It is useful because it potentially allows for the creation of stable genetically-engineered organisms. The biosafety concerns of genetically modified organisms come from the various mechanisms by which recombinant dna can "escape" and get out into the environment. An organism like this will be genetically isolated and therefore should mitigate many of those concerns. It also lessens the likelihood of further mutation over time, which can make your possibly $millions investment worthless.