Re: Q: data modeling with inheritance

On Thu, 2009-07-02 at 19:19 -0700, Nathan Boley wrote:

> Is an association, for example, an experiment that establishes a
> dependent relationship? So could there be multiple associations
> between variant and phenotype?

Exactly. You might have one group say that allele X "causes" some trait,
whereas another group might report a more precise increase in odds ratio
(for example) for the same genotype/phenotype.

> Is your concern that the number of joins will grow exponentially in
> the number of variants and phenotypes?

Not the number of joins, but the number of association subclasses. If I
have Nv variant subclasses and Np phenotype subclasses, I'd need Nv * Np
association subclasses. Multiply that by the number of association
subclasses.

> So all variants would be stored in the variants table, all phenotypes are in
> the phenotypes table, and you join through association.

Thanks. I had considered that too and that's probably what I'll end up
using.

-Reece