Wednesday, December 10, 2008

Natural variation in Nature

Nature has a whole slew of reviews on the current progress and prospects of mapping the genetic determinants of phenotypic variation:
Association mapping in humans (here)
Mapping behavioral traits in mouse (here)
Genetic variation in malaria (here)
Mapping in plants (here)
and how we should use natural variation to learn about biological systems (here)
quite a fun reading list.

Saturday, December 6, 2008

Using admixed populations to separate cis and trans effects

A new article by Price et al. looks at the effects of cis and trans acting variation on gene expression. A number of studies have approached the genetics of gene expression in humans by doing GWA mapping of the genetic determinants of the expression of a gene (usually many genes measured on a microarray). However, this approach is strongly biased towards finding signals in cis. The cis region around a gene represents a small fraction of the SNPs in the genome, and so testing for cis effects suffers a much smaller multiple testing correction than trans variation, which must be significant beyond a multiple testing threshold for the entire genome.

Price et al. cleverly circumvent this by looking in the recently admixed African American population. African Americans have on average 20% European ancestry and 80% African ancestry. Because this admixture is recent, there are only a few generations of recombination and so the genome of an African American can be thought of as a mosaic made up of big blocks of alternating African and European ancestry. So at any location in the genome African Americans differ in whether they have locally inherited 0, 1 or 2 chromosomes from African ancestors. African Americans also vary in their genome-wide admixture proportion. Price et al. use this fact, to look at trans effects ancestry, by looking at the correlation between genome-wide ancestry proportion and the expression level of genes. They contrast this to the effect of cis ancestry (i.e. 0,1, or 2 African alleles at a site), to obtain an estimate of the variance explained by cis and trans effects. Somewhat surprisingly (at least if you read mostly human genetics papers) they find that only 12% of heritable variation in gene expression level is explained by cis effects. This kind of result has also been seen in Drosophila where trans effects make up the bulk of within species variation, but contribute less to between species differences (see evolgens post on this topic: Slightly Deleterious in Trans).

See also Gene expression