Lets say one day we get to the point where we understand DNA, and how genes interact so completely that it is considered to be a programming language, Turing complete. We even develop a function library etc.
Would it be possible to then develop a virus that is undetectable due to it having almost no effect on those it infects, except for one individual.
Can we program it to go on spreading itself from person to person, say through a similar mechanism to a very mild cold, all the while testing it’s host DNA. Looking for it’s target. Once it finds it, it then “executes a procedure” to eliminate the host. Maybe it emulates Ebola or something horrible like that. So then all we need to do it get hold of some DNA of our intended target, presumably a well guarded political figure or something, and infect the less guarded people who might come into contact with him.
Is such a thing theoretically possible, given what we understand about how DNA works?
If you want single-target precision, you’re going to need to sample a lot of DNA, and compare it against a large reference “database” in your pathogen. This is probably going to require a lot more complexity than you can get in a virus. You’d have better luck with a bacterium, but we’re much, much further away from being able to custom-design a bacterium than we are a virus. I would imagine that by the time such technology were possible, various other technologies would make it irrelevant (either by defending against this sort of attack or by providing more efficient means of assassination).
I could come up with at least a half dozen major technical hurdles off the top of my head that would make this all but impossible, at least with our current technology and understanding, but in the same breath, I don’t think it’s technically impossible.
Yes it’s possible?, sure, viruses can be produced to target specific DNA sequences.
Is it a good idea?, hell no.
-
The virus might mutate to target other gene sequences.
-
The virus might also target an unrelated sequence that you haven’t tested it on (good luck testing your bioweapon on all possible human genome variations)
-
“race” is a such a hazy concept when you get down to the genetics that unless you want to target or except a very small and genetically similar population, you are likely going to have a lot of collateral damage. Target a small, isolated and recently discovered tribe? maybe, kill all French people and no one else? not a chance.
enigmatic, you’re the first to mention race in this thread. The OP was asking about an individual target.
It’ll be a looong way until we have enough understanding of genetics and biology to be able to use it as a programming language. My guess is that level of understanding is fifty to a hundred years away, and that’s if we’re lucky. However, we can make due with a cruder understanding – genetic engineering and synthetic biology can produce some amazing (if crude) things, but it’s still more of an art than an exact science.
That said, there are a lot of interesting things in the genetic engineering toolkit. Through evolutionary mechanisms, we can engineer proteins that can recognize and bind to arbitrary DNA sequences. Furthermore, we can link useful functional bits to those proteins, such that we can cut the DNA strand at a particular sequence, or initiate transcription.
Conceptually, you could use those sorts of proteins to identify sequences specific to one particular individual. But you’d need a lot of information – something like the entire DNA sequence of your target, good reference sequence (which we pretty much have now), and extensive knowledge of the range of genetic variations among humans. But since these proteins identify short (18 base pair) sequences, you’d have to use lots of them to make sure your virus was specific enough. You’d have to somehow set up a genetic circuit which activates the “kill” program only if all of the identifying proteins find their target.
That circuit, in turn, is conceptually possible, but way beyond our current abilities. Synthetic biologists are happy if they can integrate two or three signals to drive something crude like an oscillator. And you’ll probably end up with something much too big to pack into a virus.
Also, in all likelihood your engineered virus would mutate and become harmless in the wild. All that excess DNA that you packed in to it would just be a hindrance as it infects non-targets. Even if you could somehow force it to carry that baggage around long enough, that functional DNA would be subject to all sorts of neutral mutations that would probably render it useless by the time it infects the target. Or, worse, changes the virus’s specificity.
ETA: In short, such a virus is conceivable at some time in the future. But it’d be a product of research on the scale of the Manhattan project. Much faster, easier, and cheaper to use traditional ways of assassinating a target.
Impossible? “Nothing is impossible. Not if you can imagine it. That’s what being a scientist is all about.” - Farnsworth
A virus can’t read an organism’s entire genome, which is what you’d need to have a virus that would affect one and only one organism. Viral genomes are many orders of magnitude simpler than the organisms they infect, even if we’re talking about prokaryotes. Humans have 9 billion base pairs and the largest known viral genome is something like 1 million base pairs. More typical is something like 5000 base pairs.
So your program size is limited to what can fit on the medium. If the target system has a floppy drive as it’s weak point, your virus has to be small enough to fit onto a single floppy drive. If the virus is limited to 1.4 MB, it can’t store a 9 GB key, even with fancy compression. And if it can’t store the 9 GB key, it can’t differentiate between targets.
There’s no need to sample the entire genome. Current forensic analysis uses only about six alleles (I think) to uniquely identify someone from a DNA sample. The point still stands about this entire process being sufficiently complex that it would be hard to encode all this circuitry into a sub 1 Mb genome. lazybratsche, I think, came up with several good reasons why this would be technically infeasible even if the original virus could be made.
Six alleles? Assuming 2 versions of each allele, that’s only 2^6 permutations, or 1 in 64. That’s not enough to uniquely identify someone. It sounds to me like that’s a standard number they use to exclude someone, as in a paternity test. But it sure isn’t enough to uniquely identify someone out of a population of 6 billion. And note that 1 Mb genome is the largest viral genome known, most are much much smaller. HIV is only 10 Kb. Wait, it’s actually more than 10Kb, because it has 10,000 base pairs, but the genome is base 4.
It is impossible for a virus to read someone’s entire genome to uniquely identify someone. Viruses just don’t do that–they just hijack the molecular machinery of the host cells to make more viruses. They don’t have any molecular machinery to allow them to read the genome of the host.
The loci used in forensics are usually short tandem repeats (STRs), not single nucleotide polymorphisms (SNPs). At a given STR, there are essentially many alleles (as many as several dozen in some examples).
Furthermore, each person has two copies of each loci, and each copy can be considered unique. And the US standard in DNA profiling uses 13 such loci. So if each loci has an average of 10 variants*, you actually end up with 10^26 permutations. That’s reduced by the fact that certain alleles are much more common than others, but it’s still a lot, and enough to uniquely identify a person for most practical purposes.
However, off the top of my head I can’t think of a nice virus-packable mechanism that can identify that sort of allele. A protein can identify a sequence, but it can’t count how long a stretch of repeats is. And DNA binding proteins get real sloppy when they target repeating sequences…
My idea was to identify a number of rare SNPs in the target. Everyone should have a few, just from random mutations. If you can reliably find SNPs that occur at a frequency of 1 in 100, then it would only take five to uniquely identify your target from a population of ten billion. That task of identification is actually the easiest thing out of the whole concept. We can already fully sequence a human genome for something like $10k (cheap by the standards of biomedical research). And there are lots of people that are sequencing larger and larger samples of the population, so that we can better understand natural genetic variation.
*I’m making that number up, but it should be in the ballpark.
So are you saying we will have to test the entire DNA sequence or not?
If not, are you saying we can possibly fit enough “code” onto the medium without causing too much instability?
The virus’s designers would need the entire sequence, yes. With the entire genome, they could identify a small number of mutations specific to the target individual. They then engineer a virus that will recognize that handful of specific sequences.
Maybe? Hard to say for sure.
Let’s say my scheme uses eight genes, each which identifies a rare mutation in our target. And, since it’s my pretend virus and I’ll do what I want, let’s say that each of those targeting genes will simply release a transcription factor, so that the “kill” genes are only transcribed when all eight transcription factors are free. That’s sixteen genes so far. We’ll probably need some additional regulatory genes for good measure. Plus a handful of kill genes… call it 30 genes in total. The total sequence would probably be around 60 kilobases in length.
Now, compared to the genomes of common viruses, that’s a huge payload. A typical rhinovirus genome is around 8 kilobases. The herpes virus has one of the largest genomes for a common virus at up to 230 kilobases, so it might be a reasonable vector.
But even then, viruses reproduce quickly and mutate rapidly. I doubt you could count on the virus spreading far and wide while still retaining the specificity and deadliness you want. You might have some luck just introducing the virus into a public place where you can reliably find your target. That way the virus doesn’t have to spread from person to person before finding the target - it just infects a few hundred people in a public space, but only kills the target.
But there are a bajillion technical hurdles and opportunities for things to go horribly wrong, so it’s still way easier to just poison your target by traditional means.
I’d put this scheme in the same league as a space elevator - sure it’s possible, but it’s going to be difficult and require some pretty big technological leaps.
Although I must say that the whole “artificial life” thing may happen sooner than we (or I) expect - google Craig Venter…
Yeah, I actually tried to edit the original post, but took too long and didn’t have time to make another post so apologies to the OP, was thinking about the context this question usually comes up in.
A single individual probably makes the science easier, but you are still talking about something on the order of the human genome project (at the least) to kill one target, most would be assassins are likely to invest the money in bombs.
If you wanted to kill everyone who was carrying some specific genetic marker, like, say, you have a pathological hatred of everyone who was likely to get deep vein thrombosis from their Factor V Leiden mutation, that would simplify matters somewhat.
Somewhat.
And if you could get hold of an entire DNA profile it might allow you to identify possible alternative strategies. For example, certain metabolic enzyme expression differences can make individuals much more vulnerable to some kinds of poisoning.
One other problem with targeting a single individual, is that depending on the situation, your innocent bystanders might be the most likely to share the targets genetic profile, I’m assuming you don’t want to kill the targets kids.
I know this is possible, because it happens in Metal Gear Solid
How about the premise of Frank Herbert’s The White Plague: a disease engineered to be harmless to men, but deadly to women? Would that be easier than targeting one person? Easier enough that it’s close to state of the art and needs a place in my closet of anxieties along with the Canary Islands tsunami, Yellowstone exploding, and the Earth farting us to death?
Actually, it’d be easier to come up with a disease that affects males but not females. Males carry the Y chromosome, but females do not. Females “just” have an extra inactive copy of the X chromosome. It’d be easier and much more reliable to make your engineered disease target some bit of the Y chromosome, rather than try to determine the Y chromosome’s absence or count the number of X chromosomes.
If our disease engineers are happy with sloppier discrimination between males and females, they could design the pathogen to sense various sex hormone levels. For instance, you could have the disease only kill off hosts with high testosterone levels. That would disproportionately affect males, but it would also affect females with unusually high testosterone, while not affecting males with unusually low testosterone. If you picked a female sex hormone, you’d have similar difficulties, plus the fact that female sex hormones vary dramatically over the menstrual cycle. So you might end up with a pathogen that kills women at a certain point in their cycle (and males with unusual hormone levels), while not harming post-menopausal women.