Genetic Variance Between Pops as compared Within Pops

1. Sep 12, 2007

Simfish

As it often said, the majority of the variation in the human genome is due to intra-group variation, not inter-group variation (I think .85 is the number due to intra-group variation). I know what this intuitively means, but can anyone explain to me what it rigorously means? (in statistical terms?)

2. Sep 13, 2007

EnumaElish

I don't know how group is defined, but it doesn't matter as long as YOU know.

Suppose you have a numeric index of human genome code for each individual i in group g. Let that index be Y: H --> R. That is, for genome h in H, Y(h) is a real number. Each group is a subset of H. Subscript i indexes the h's in group g; i = 1, ..., Ig where Ig is the number of h's in group g. There are k groups.

A regression equation of the following form is estimated:

yi,j = a + b1d1 + ... + bkdk + ui,j

where the dependent variable is Y(hi,j) = yi,j and the independent variables are the d's; u is the error term. For each g, dg = 1 if j = g (that is, hi,j is in group g), dg = 0 otherwise. The R^2 statistic of the estimated equation is expected to be 0.15 (a poor fit).

Last edited: Sep 13, 2007