clearer graphic of using DNA to solve TSP ... it's starting to make sense even to me
Wow. I think by slavishly redrawing the last post's filched graphic, I am dimly beginning to comprehend how Leonard Adleman used a Petrie dish of wet DNA strands to compute the shortest path of a 7-node Travelling Salesman Problem (TSP).
In Nature, the information stored as a sequence of the nucleotides
T = thymine
C = cytosine
A = adenine
in a DNA strand eventually acts as instructions to create particular amino acids. These float free in a chain, and then chemical attractive forces "fold" the chain into a large 3D solid protein.
Here, Adleman isn't interested in amino acid or protein manufacture.
He just wants to use DNA's computational machinery -- which in Nature corrects errors during replication -- on short DNA strands (oligonucleotides) which store digital data in a code designed to represent the 7! = 5040 different possible paths in a 7-node TSP.
DNA sequences are left-right ordered; the order matters; in the proper order you get a tree frog or a tulip; in the wrong order you get gibberish. 5' and 3' are the designations of the two unique ends of a DNA strand.
Anyway, how hard can this be?
And is there a future for goopy DNA Petrie dish computing?
Or, as we speak, is the National Security Agency running a secret biotech laboratory 11 storeys underground using DNA to crack Russia's, China's and Iran's secret intelligence and military codes?
The buzz is that Adleman's technique does these massive amounts of digital computations and gets the answer in blinding speed. At least in very short order you get goop in a Petrie dish which represents the answer. How you extract your answer from the goop in the Petrie dish -- this might be a little more time-consuming and difficult.
But biotech labs have been doing all these things routinely for about 20 years, in a fairly straightforward, mechanized, automated, computer-controlled way, particularly since Kary Banks Mullis's discovery of the Polymerase Chain Reaction (PCR).