[extropy-chat] Why no assembler design?

[Eugen Leitl]

On Tue, Nov 18, 2003 at 10:59:32AM +0100, Alfio Puglisi wrote:

> I'm sure that the question has been debated to death somewhere else, but
> if the components are small and simple enough, could they be produced
> biotech-style (i.e., assembling some billions of identical molecules in a
> dish via chemical reactions), and then the assembler built combining the
> basic blocks? Or is this an unworkable approach?

In machine-phase that would be convergent assembly

	http://www.zyvex.com/nanotech/convergent.html

It is very possible to design small, very stable proteins (it is relatively
easy to design extremely stable proteins with site-directed mutagenesis.
Basically, you have to tesselate a space (say, an elementary cell of a
crystal) into 3d jigsaw, and find out the sequence for each jigsaw piece (the
inverse protein problem).

In practice, it's so tangled one would tend to just mutate random pieces of
proteins, fold them individually (including forming complexes with functional
groups -- as SWNT can be solvatised with SDS it should be also possible to
solvate and assemble them with proteins) via a protein folding code
(selecting for fast folders), and then let assemble the resulting soup of lego
blocks.

Extra bonus points for crosslinking individual jigsaw pieces, as well as cell
contents; either by using natural amino acids, or use artificial amino acid
analogs.
 
> Blocks like that could have unique markers on them, allowing the parts to
> snap together in the right way. To accomodate for this, the design would

Complementary surfaces are sufficient, actually.

> be complicated, and size increased to maybe 5x or 10x. Still, it wouldn't
> be a bad first step.
> 
> Could it be possible to design an assembler using parts that can snap
> together using unique marks, accepting maybe a 10x increase in size? It
> wouldn't be a bad first step.
> 
> I suspect one needs a well-developed nano at home system just to
> contemplate this scenario :-))

It takes a Blue Gene type of system to deliver enough performance for folding
small fast folders via brute-force MD (the only method which is guaranteed to
work, because it simulates what the molecule actually does in the solution).

You can't scale that over a p2p grid. You can sample lots of conformation
space, though, and you certainly can stochastically-driven combinatorics in
design space. Both are embarassingly parallel enough to only require
communication each few hours or days.

-- Eugen* Leitl <a href="http://leitl.org">leitl</a>
______________________________________________________________
ICBM: 48.07078, 11.61144            http://www.leitl.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE
http://moleculardevices.org         http://nanomachines.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20031118/dc514dc1/attachment.bin>