Infection of biological DNA with digital Computer Code | Hacker News Reader

Infection of biological DNA with digital Computer Code

pastebin.com

36 pointskyberias40 comments

This is pretty trivial - of course we can encode anything we want in DNA - people have done cool things with this since synthesis was first invented. And assuming DNA synthesis technology continues to improve rapidly we might even be able to synthesise bacterial genomes routinely in the next 10 years.

So you can write a computer program to a bacterium. Then you can sequence it and the code goes back to being stored on a computer.

But it ends there - the code never gets executed again. FASTA files are just text files that would rarely have executable permissions and never actually be executed by a user. Most boring and uneventful outbreak ever.

Not boring or uneventful at all! An amazing way of transmitting information across eons or under in regimes of restricted communication!

If I can infect a common fly with my stenographed, encrypted message, then I can communicate with people in China! I can translate our major books in every language and hide that data in cockroaches! If we ever nuke ourselves so badly that we only have people living in bunkers, then maybe one day they will discover that we hid all of our knowledge in life! Maybe this has already happened! Maybe aliens did so with mathematical proofs when they discovered our world an eternity ago!

I reject your claim of boring! THIS IS AMAZING!

Angersock's 1st Rule of Storage:

"Any sufficiently advanced compression or encoding is indistinguishable from /dev/null"

I would choose /dev/urandom. But you might as well be writing to /dev/null.

Edit: beat by 3 minutes and a stale cached page.

Or maybe /dev/random ?

In a "we just nuked ourselves" scenario, integrity of genetic information is one of the first things to go.

AMAZING enthusiasm! :)

Right now I can go to one of many websites to encode a program into a DNA sequence[0] and then to another website to paste that sequence into a text box and order that DNA.[1] This is all doable right now by every one of us. It's just not useful yet.

[0]http://www.sgidna.com/translator.php [1]http://dna.macrogen.com/eng/service/seq/standard/standardseq...

edit: Let me add that I think this is crazy interesting and even if the DNA is never executable itself, it's a very very interesting way to store a program that can later be extracted.

I was at Compute Midwest (a tech conference in Kansas City, Missouri) this past fall and saw a demo of this very thing by Autodesk's Andrew Hessel. He ordered some customized bacteria (maybe yeast? I don't exactly recall) with the conference title encoded and then sequenced into the microbes. It was very interesting to me that that level of manipulation is available to any of us, right now (even if it's not exactly useful).

It did bring up some concerns, though -- do we now have mail-order anthrax manufacturing? This isn't something I'm seriously worried about, but it does seem like something to consider in the future as this technology evolves.

Something else I just thought of while typing this is that this would make an awesomely sci-fi means of information storage. No need to remember a password, we've encoded your gut flora with your private key!

>Something else I just thought of while typing this is that this would make an awesomely sci-fi means of information storage. No need to remember a password, we've encoded your gut flora with your private key!

Was the first thing that occurred to me too. Talk about steganography.

First occurred to me as interesting too. Then I realized that you'd be dropping your private key in every toilet you defecate into. Not to say that secure intra-body information storage isn't possible though.

Then again, re: the steganography -- gut flora's pretty diverse (and there's a lot of it). People might not necessarily think to, or want to, dig through your shit.

I love the internet. I was at Compute Midwest too. His talk was where I initially learned about this. Are you from KC?

Yep, sure am! I really like CM as it's a "big ideas" type-of tech conference, compared to the other sorts we tend to have in the area, which are more geared towards flavor-of-the-month web framework demos and the like.

I doubt the author has any comprehension of the errors involved in synthesis of a physical sequence from a text string, the mutations introduced into the sequence during the lifecycle of the organism, or the errors involved in sequencing the DNA back into an abstract text string. Life is largely accustomed to these errors and drifts of sequence, computer code is incredibly intolerant.

A serious discussion of this concept would involve encoding the information with some sort of channel coding and calculating the likelihood of any such program of minimal length successfully propagating itself.

I'm not sure the validity of the post or the information therein. It did lead me to think of some futuristic use of implanting data into DNA:

Could it be possible for someone to take a corporate secret or information, translate it into DNA, incorporate the DNA into a virus, and then infect themselves with said virus? While they are incubating say, the common cold, they could be in physical transit to their destination with no electronic device. The virus could ultimately be extracted and it's information sold to the highest bidder.

Actually, it sort of reminds me of a novel by Alan Dean Foster - Sagramanda http://www.amazon.com/Sagramanda-A-Novel-Near-Future-India/d...

This would be possible but not a good data storage strategy - high mutation rates would lead to very rapid data degradation. A workaround would be to design a highly redundant encoding so the probability of sequence modifications altering the encoded information was low.

No it doesn't work like that, mutation isn't uniformly high, some sectors are very stable.

It does work like that: in bacteria (as in the OP), there are no highly stable intergenic regions, and in viruses (as in grandparent's suggestion) where there are rarely significant intergenic regions at all, the mutation rates are the highest recorded.

Wait, yes it does! Mutation rates may be small over generations due to SELECTION. We're talking about nonsense DNA that does not have any purifying selection pressure, ie. it does not encode any useful or critical protein for the organism.

But some segments of DNA are effectively unchanging and you can place the message in those segments "junk" sub-segments.

And that, ladies and gentlemen, is how the zombie apocalypse started...

Anyway, those areas you guys are calling non-coding and "junk" do a lot more than we understand, and I would venture modifying them willy nilly could have some pretty nasty outcomes.

Actually, we're beginning to understand them more and more. Remember, if it was as easy as a mutation here or there then the zombie apocalypse would have already started. I'm much more afraid of grey goo (look it up) than punching out some TAGACAs.

On the ZA, who says mutation is the mechanism, maybe someone wants the ZA to start....

On non-coding regions, we are indeed learning more and more, but these developments are relatively recent and there is much more to be discovered.

As for the grey goo... well, that would be quite the singularity, wouldn't it?

Creating an artifical bacterial genome and transplanting it into a bacterium is still a huge effort and far from routine. As far as I know it has only been done once, and it did cost some tens of millions of dollars.

Genetic manipulation is typically done at a much smaller scale, and not by exchanging an entire genome. And on a small scale it would be very unlikely that such a switch would go unnoticed.

The way back from bacterium to executable is also extremely unlikely, there would be enough mutations after some generations that the program is unlikely to run at all.

And next thing you know, the Melding Plague:

http://en.wikipedia.org/wiki/Technology_in_Revelation_Space#...

Yikes. That was one of the most horrifying parts of the series.

"developed an encoding technique from DNA to human letters"

Should have used UTF-8. Also human letters does not equal greco-roman capital letters. Wake me when they embed some kanji.

All you need to encode ANYTHING in DNA is 2 or more symbols. IF nothing else you encode whatever data you want into binary.

kanji -> UTF-8 -> Binary -> DNA.

Not buying it. The point of the article was coding roman letters. Its a "cheat" to claim it supports UTF-8 klingon glyphs because technically its possible to write the string:

"U PLUS SIGN F EIGHT E FOUR" is the glyph for Klingon TLH.

Or more likely given the limited codespace its actually:

"U PLUS SIGN F VIII E IV" is the glyph for Klingon TLH

"real" unicode support would look like an imaginary packed UTF-5bit rather than existing UTF-8 or a roman letter prose description of UTF-8

As a meta discussion, much as being able to encode arbitrary text including source and executables "proves" that bio-mechanical DNA virii are possible, I guess this post where Klingon glyphs can be described in someone elses roman letter glyphs in earth DNA also "proves" that inter species human klingon romance and reproduction is viable (I'm sure to the relief of Lt Paris and Lt B'Lenna).

(edited to add because of codon degeneracy I'm claiming we should only pack 5 binary bits per codon, I know there's 6 theoretical bits per codon and only (about) 4 bits of amino acid per codon so I split the difference and went for 5. Its an interesting puzzle, do you allow any codon even if its biologically nonsense or are you writing a distinct amino acid alphabet using the codon as a representation of the amino acids?)

This is an interesting, albeit somewhat scary, application of Automaton Theory.

Scientists shouldn't be rushing with this kind of thing (research for sake of research), and we, as humanity, should carefully consider the implication of this kind of things.

Btw, this reminds me of the old game called Core Wars (http://en.wikipedia.org/wiki/Core_War).

My analysis is that it looks plausible and not really surprising if one knows a little bit about molecular biology. The extension of this is that one could find a vulnerability (buffer overflow maybe) in one of the software used by gene analysts that can be used by the injected DNA to infiltrate a PC.

It would be extremely surprising if any exploit could be triggered by a particular valid DNA sequence (which it would have to be to be able to be sequenced/synthesised).

Interesting, now I wonder if a malicious physical virus could form a digital-physical symbiosis with an artificial intelligence and be able to evolve itself through DNA synthesis with the said method and eventually wipe humans and human controlled computers out of the game.

I'd be more worried about someone taking a known sequence like modified HN51 or smallpox as the payload. Next, combining it with effective malware that targets FASTA files. One of those "incautious scientists" thinks they are synthesizing something benign, not taking precautions, and a highly infectious disease pops out.

Assuming that's not how human minds originally evolved in the first place. Its a big world out there, if there's a tasty niche its probably already filled.

Is this April's fools day?

Never imagined that someone else could love X-Files as much as I do.))

Is this real? If someone isn't rolling on the floor of their creepy, poorly illuminated lab, laughing; this is huge!

This is a theoretical concept based on real data and experiments.

nice project in python to encode binary files to DNA https://github.com/Allanino/DNA

Reply on news.ycombinator.com