Fun_People Archive
26 Jul
Making DNA Add

Content-Type: text/plain
Mime-Version: 1.0 (NeXT Mail 3.3 v118.2)
From: Peter Langston <psl>
Date: Fri, 26 Jul 96 22:23:45 -0700
To: Fun_People
Subject: Making DNA Add

Forwarded-by: Keith Bostic <bostic@bsdi.com>
Forwarded-by: "Daniel V. Klein" <dvk@lonewolf.com>

"Making DNA Add", Frank Guarnieri, Makikio Fliss, Carter Bancroft
	    Science, Vol 273, 12 July 1996, pp 220-223

Abstract: "Recent studies have demonstrated the feasability of using
DNA-based experiments to compute solutions to combinatorial problems.
However, a prerequisite for designing a computer useful in a wide rane of
applications is the ability to perform mathematical calculations.  The
development of a DNA-based algorithm for addition is presented.  The DNA
representation of two nonnegative binary numbers is presented in a form
permitting a chain of primer extension reactions to carry out the addition
operation.  To demonstrate the feasibility of this algorithm, a simple
example was executed biochemically."

The remainder of the paper details the algorithm and results, and goes on
to say that "The first generation algorithm described here has an obvious
limitation.  Because the output is encoded in a different form from the
input, it is not presently possible to perform either iterative or
parallel addition.  The development of a more mature DNA-based algorithm
will require modification of the present procedure to take full advantage
of the enormous potential of DNA to engage in massively parallel
reactions."

And further "the algorith describe here is not technically demanding,
because the simple biochemical procedures involved require approximately
1 or 2 days of laboratory work."

So, while the integer calculations are still measurable in IMPS (that's
Instructions per Million Seconds, 5-10 IMPS in this case), I am
nonetheless reminded by the 3-page foldout illustration (p65) of a
parallel multiplier from "Automatic Digital Computers" by Maurice Wilkes
(John Wiley & Sons, 1956), where certain technical problems in programming
are noted, or where a description of a magnetic drum made by Ferranti Ltd
is described (p181), where it is claimed to hold 128 words of 20 binary
digits on each of 256 tracks, the bit density being 98 bpi.  Given another
40 years or so, we may have real biochemical computers...

© 1996 Peter Langston