On 23 Feb 1995, Anders Sandberg wrote:
> Peter Merel wrote:
> >A couple of years ago I proposed a scheme to prevent _all_ existing viral
> >diseases from using humans as a vector. The idea is fairly simple; we encrypt
> >human DNA with a computationally inexpensive cipher, and build the decryption
> >algorithm into the ribosome. Since viral DNA will not be replicated by such
> >a ribosome we should be able to give ourselves several million years breathing
> >space until new viruses are able to evolve.
>> I see a problem in the otherwise sound idea: dormant retroviruses. It is known
It SOUNDS good but it cannot be done. There are already multiple levels
of protections built in, both from mutations of various sorts AND to
foreign genetic material. Nevertheless, mutations get by the check,
recheck, and repair machinery to the tune of one base misincorporation
per 10^7 replication events. That is the OBSERVED average error rate.
It is higher in the less transcribed regions of the genome.
As for viruses, they have and will continue to evolve runs around our
defenses. Free, nonchromosomal DNA isn't "liked much, and neither is
double-stranded RNA. The cell will react to both by degrading them and
turning on an "antiviral state" via interferon, for one. Even so,
viruses develop end runs around these defenses either by replicating
quickly or by loading the cell with short RNA fragments which reverse
interferons activation. Very cagey.
The key problem with the encryption idea is that it is based on a false
idea of genetic and protein stability. Rather than static structures,
they are very dynamic. This combined with the fact that chemical
reactions such as take place to produce nucleotides, proteins, duplex
DNA, RNA, etc are ALL subject to "error" by very nature of chemical
reactions. Side reactions, back reactions, are all inevitable and have
an effect of preventing stasis of structure. Proteins (and RNAs), which
do all the work, "breath," change conformational state, can be
mis-translated or transcribed, etc. They cannot be made impervious to
error. Nothing can.
In any case, at the level of the DNA molecule and the bases that make
it up, all DNA looks exactly the same. One creature's nucleic acids
cannot be delineated from another, though methylation goes in that
general direction...yet it is a chemical process and error prone - and
The kinetics of viral infections ensures that there will never be a
sure defense from viral infection. They replicate like sons-of-bitches,
and they tend to be mutators. The fact that they can also serve to
mutate cellular genes too makes any genetic defense ultimately subject to
mutation and inactivation.