First Reproducing Artificial Virus Created

jrrl writes "USAToday is reporting that Craig Venter's research group has synthesized a virus from scratch and that it "became bioactive" (started reproducing). Particularly interesting is that it only took them two weeks to build, rather than several years that previous attempts had taken."

Comments

What they did, why it is hard

[sam_handelman]

The human genome (which is DNA), contained in each of your cells, contains the instructions needed to make a cell (much like a computer program.)

However, in order to use these instructions to make a cell, you need a cell of the same kind to read them.

Analogy: You have a computer program that tells you how to manufacture computers but this doesn't do any good unless you already have a computer OF THE SAME KIND on which to execute it.

So, even if I assemble an entire human genome, I can't use it to make a person unless I already have a human cell. Kapish?

A VIRUS, which is what was made here, is NOT A CELL. It is a parasitic piece of DNA that hijacks an existing cell and contains the instructions to make viruses. The DNA that the virus contains is, in the best case, sufficient to hijack the cell all by itself, and convert the cell into a factory for making viruses. Viruses CANNOT make more viruses by themselves. The similarity to a computer virus, I assume, is obvious.

So, if you can make VIRAL DNA, this will be sufficient to make the virus, if you have cells that the virus can infect.

Even making the genome of a virus is very difficult. The "commercially available" DNA mentioned in the article is made chemically. DNA is made up of a chain of monomers; each monomer has a 5' end and a 3' end that can attach together to form a chain. In order to add monomer n+1 to a growing chain, this is what you do (description meant to be accessible to people who don't know a lot of chemistry): ...(Monomer n-1) 3' - 5' (Monomer n) 3'(BLOCKED)
-> **add reagent to unblock**
-> wash ...(Monomer n) 3'
-> add 5' (Monomer n) 3' {BLOCKED}
-> add reagent to attach 5' and 3' together ...(Monomer n) 3' - 5' (Monomer n+1) 3' {BLOCKED}
and repeat for Monomer n+2. Recursion is good.

Now, this is done in parallel in thousands of molecules of DNA (the 5' end of each molecule is fixed to a plate.)

Every time you add the reagent to remove the BLOCKS, it has a percentage chance, which can be very small, of failing.

So, for example, if, on one paritcular molecule, it fails at position 10, then instead of:
ACGTACGTACGT
you will get,
ACGTACGTAGT.

DNA that makes proteins has something called a "reading frame", consisting of codons which are three monomers long. If you shift the reading frame over by 1 monomer, it completely changes the meaning of the message.

So, a single nucleotide deletion, which I describe above, is disastrous - the synthetic DNA becomes useless.

Even if the chance of failing to remove a block is small - typically about 0.1% - if your DNA molecule is thousands of bases long, the chance of successfully adding every base to any individual molecule is slight.

Of course, you can make two different 100-base long molecules by the above technique and then ligate them together (recursion by splitting the task in half) which is, I believe, what's been done here. This has technical difficulties of it's own, of course, but with refinements it woud allow you to make useful DNA of length n*2^m instead of DNA of length n.

This is a frightening prospect because it would allow you to make ebola "from scratch", or just from the the string of letters that represent the genome (which is so short I could write it out by hand on a stack of cocktail napkins.) We're not to that point yet but it is a scary possibility.