Re: On Human Diversity From The Scientist
- From: "JAE" <jae@xxxxxxxxxxx>
- Date: 26 Oct 2005 11:43:40 -0700
rmacfarl wrote:
> JAE wrote:
> > Jois wrote:
> > > http://www.the-scientist.com/2005/10/24/16/1
> > >
> > > Suggests why some areas of human diversity get studied and some don't.
> >
> > The author suggests three reasons why traits like height get less
> > attention than disease traits. I think he underestimates the problems
> > associated with #3--that the former class of variation are products of
> > multiple loci interacting with environment in often
> > less-than-predictable ways. It's curious that he proposes mapping by
> > admixture linkage disequilibrium
>
> Could you please explain what this phrase means Jason?
Briefly, mapping by ALD (ALD mapping or MALD) is a technique whereby
you look at large blocks of DNA in a population recently formed from
the mixing of previously long-separated populations. In the new admixed
population, some individuals will possess traits, others will not
according to the different DNA that they inherited.
We don't generally have the luxury of first looking and genes and
figuring out what they do. What usually happens is that we look for
variation between individuals (e.g. a particular SNP--perhaps
non-coding) and then see if this variation co-varies with some
phenotypic variation. If yes, then we can start looking around the SNP
for candidate genes that got inherited with the possibly neutral SNP,
then go back and see if the genes similarly co-vary. This is difficult
though, as recombination shuffles about chromosomes rather quickly such
that unless the SNP is right next to the gene, something that doesn't
have to happen, after not too long, recombination will put the SNPs in
equilibrium with any variation elsewhere and they'll be useless to
identify any other variation. The solution *may* be to look at more
markers, but there's no guarantee that there will be this type of
variation right next to a functional gene.
Imagine this chromosome in a person.
----------X---------C-----------
where big X is a gene and C is a neutral marker that we've discovered.
The capital X may cause a disease (or blue eyes or a penchant for being
really tall, etc.) in this but we didn't find that. We found a C at
some SNP elsewhere on the chromosome. Little x doesn't cause the
disease and T is the alternative marker at the other chromosomal
positions.
e.g.
----------x---------T-----------
Now there are also chromosomes:
----------X---------T----------- (the disease, but different
associated marker)
and
----------x---------C-----------
all of which can be generated by recombination. After not too long,
all of them do exist and as such, the X-x polymorphism (our unknown) is
in equilibrium with the T-C polymorphism and one isn't useful for
identifying the other. We won't find any co-variation to help us
narrow down the search for X.
We can look at a whole lot more markers and hope that there's something
like this,
----------X-G-------C/T-----------
Where the G is so close to our gene that recombination shuffles it less
often, but that's rarer and even so, eventually, recombination will
break this up too. It just takes longer, but it will happen.
MALD is one way of getting around this problem. The admixed population
will have a subset of the variability and early on, recombination won't
be as big a deal. In a recently admixed population, the chromosomal
blocks haven't undergone as much recombination and as such, markers
farther away may still be linked to genes. A natural bottleneck
filters out only part of the neutral variation and it's early on linked
to the polymorphic gene-state of interest.
We hope that in this population, we just got the
----------X---------C-----------
type from one population and before recombination we'll get one or the
other chromsome block in people and have clearer association between
the variability we've found and the disease.
MALD looks at several sites on several chromosomes (several 100 to
several 1000s) and looks for association rather than looking at
millions of sites. It's just a way of improving the odds. It's a very
useful technique. It doesn't always work, but it's helpful in many
instances.
[Looking back, I don't think I explained that very well, but I hope
it's a start.]
> > as a means to start addressing some of
> > this variation, but on the success stories, falls back on genes
> > identified with hypertension rather than some other non-disease aspect
> > of biological variation.
>
> I wonder if that's because there is more data available on genes that
> are linked to disease, especially one as common as hypertension?
More data is available because diseases get charted when people get
treatment. There's also more discrete diseases (e.g. you have it or you
don't) rather than more continuous traits like height. More funding is
available for looking at this too. All are important.
.
- Follow-Ups:
- Re: On Human Diversity From The Scientist
- From: rmacfarl
- Re: On Human Diversity From The Scientist
- References:
- Re: On Human Diversity From The Scientist
- From: JAE
- Re: On Human Diversity From The Scientist
- From: rmacfarl
- Re: On Human Diversity From The Scientist
- Prev by Date: Re: Early Hominids and Dry-Season Adaptations
- Next by Date: Re: The Goldilocks AAT Theory Re: Gorrillas use tools, too
- Previous by thread: Re: On Human Diversity From The Scientist
- Next by thread: Re: On Human Diversity From The Scientist
- Index(es):
Relevant Pages
|
Loading
