[Gllug] Biometrics?

[Nix]

On 23 Mar 2009, John G. Walker stated:
> There are approximately 25,000 "genes" (in the sense of strips of DNA
> that create proteins) in the human genome. Add too this the vast amount
> of "junk" DNA (which may not be junk, but is certainly individual), and
> then try to estimate how long it would take to record all that
> information from a single individual.

Well, that's the thing, you see. None of this is truly 'individual'.
What DNA fingerprinting does is records the length of a bunch of tandem
repeats (these days, short tandem repeats) in each individual. Not many
are recorded: in the US it's 13 and in the UK I think it's 15 or 20.
These areas tend to mutate (specifically, lengthen) fairly frequently
(but 'fairly' is very much a relative thing: most STRs are inherited
unchanged at any one time).

The STRs chosen are scattered across the chromosomes (IIRC, none are on
the X chromosome to avoid difficulties with Barr bodies and the
intrinsic mosaicism of all women). So linkage is low and you'll have a
roughly random selection of your parents' fingerprinted STRs, perhaps
with one or two lengthened a bit.

But there's a problem, even ignoring the presence of identical
twins. Tandem repeats are, by definition, repetitive regions where DNA
replication tend to lose its grip and accidentally lengthen on the fly.
Nothing constrains this to happening in the germ line (although the male
germ line, at one replication every 14 days, does tend to accumulate a
lot of mutations of all kinds). There will certainly be cases out there
of individuals who have different lengths of fingerprinted tandem
repeats in different parts of their bodies: fingerprinting would, I
suppose, see such individuals as two or more different people.
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug