Codon Optimized for Failure

STATUS: Active (last updated Endy 17:07, 17 Oct 2005 (EDT))

What if we had a genetic code, which supported translation of all 20 natural amino acids, but that any point mutation resulted in a non-sense codon? In other words, what if every point mutation resulted in a translation stop? Such a code might form the basis of a "fail-fast" genome.
To do this, it needs to be true that any point mutation maps to a stop codon. Presume that we only need twenty codons. For a three-base codon, there are 9 possible single point mutations...

AAA <- original codon
AAC [nine single point mutations...
AAG
AAT
ACA
AGA
ATA
CAA
GAA
TAA ]

Thus, starting out simplemindedly, we need 20x9 "stop/non-sense" codons (with more brain power, it should be possible to optimize the mapping into fewer stop/non-sense codons).

Since a 3x4-base codon table only encodes 64 codons... we should consider bigger codons. Peter Schultz's group has shown that 4-, 5-, and 6-base codons work. So, what about using 4-base codons. A 4x4-base codon table would encode 256 codons. For a four-base codon, there are 12 possible single point mutations...

AAAA <- original codon
AAAC [twelve single point mutation...
AAAG
AAAT
AACA
AAGA
AATA
ACAA
AGAA
ATAA
CAAA
GAAA
TAAA ]

Thus, continuing simplemindedly, we need 20x12 "stop/non-sense" codons.

Since a 4x4-base codon table encodes 256 codons, and 240 < 260, we should enough "room" in mutation space to bracket each coding codon with 12 failure codons!