Slashdot Mirror
Parity Code And DNA
jnana writes "There's an interesting article in Nature about error-correcting parity code in DNA. It seems that there are enzymes that check for even-parity nucleotides (according to a 0 and 1 assignment scheme in the article) and recognize odd-parity nucleotides as errors. The authors argue that this parity scheme is the reason that adenine, thymine, cytosine, and guanine became the building blocks of nucleotides instead of other types of purines and pyrimidines that must have coexisted with them."
Several bio-tech and software companies, the names of whom are undisclosed, today filed a joint patent infringement lawsuit against Universe Creator, Inc. [GOD, 12777.12, +65-1/8] for allegedly violating a patent on the use of parity checking algorithms in DNA genetic sequencing. "We don't see any evidence of prior art.." a legal representative commented, "Much like Corn(tm)--we invented that too. That stuff the Native Americans used to grow was some other crop they called 'maize.' We've been gyped for all these years." If they win the lawsuit, the parties involved will release an army of Red Robots to extract royalty fees from all Hu-mans.
In the meantime, I'm still searching my newly aquired Gene Map for a copy of DeCSS...
This sounds to me of another case of "If you look for something hard enough, then you'll find it". In actual fact there is no parity checking there at all.
Parity checking (in computers) involves adding up the number of 1's in a byte, and putting another bit on the end purely as a form of error control. In even parity there are always an even number of 1's in the (byte + paritybit).
In the article they've figured out that cytosine has 1 donor, and guanine has 2 donors. Then they invent the whole parity thing by letting cytosine equal 1 and guanine equal 0, and when you add all the numbers together you get an even number of 1's. Sounds like circular reasoning to me.
If there was a regular parity bit in the DNA to make sure that an even number of G's or A's occur every 8 or so pairs, then fair enough! But what the article is describing isn't the same parity as you get on your serial port.
It's like saying "Let all starfish with 5 arms equal 1, and all starfish with 6 arms equal 0. Add the numbers up and bingo we've got even parity! Nature is a computer!"
Stinks of looking for more funding to me.
I am artificially intelligent.
As the previous poster pointed out, there isn't any parity to speak of in DNA. I have no idea why they're trying to make the comparison.
Hydrogen bonding is a much higher-fidelity error correcting system than parity checking. With parity checking, you can catch any single error (the number of 1's changes), but not any pair. With hydrogen bond donors and acceptors, you have to have an exact match at all positions. Any number of errors ruins the complementarity.
For what it's worth, the base pairing system is quite elegant: due to size constraints, purines (A,G) and pyrimidines (C,T) must pair with each other. Using the article's notation of 0=H-bond acceptor, 1=donor, the four nucleotides are
A = 10_
C = 100
G = 011
T = 101
where _ is simply a hole. The best pairwise complementarity is (evidently) AT and CG.
It's a chemically implemented RAID-1 system, not a parity check.