[GenABEL-dev] faster polygenic

[Yurii Aulchenko]

William et al.,

I was running more tests and what appears is that

1) your 'null.herit' is probably the right one -- it coincides very
well with simulated one, while on average 'old' polygenic slightly
over-estimates, while polygenic_hglm slightly under-estimates

2) still, FMM generates L which are >1 (1.014), and this is VERY
significant. Old and hglm-based tend to be a bit on other side -- L <
1 (0.996 for old, 0.993 for hglm), and for hglm this is more
pronounced

Xia: previous point is still there, if I cf with William's
implementation, the picture is quite the same (bi-modality, etc) as on
graph I have sent before, though I can only cf h2 and lambdas (L), as
totalVar is not reported by FMM

Yurii

On Mon, Feb 28, 2011 at 11:44 AM, William Astle
<william.astle at imperial.ac.uk> wrote:
> On 27/02/11 21:51, Yurii Aulchenko wrote:
>>
>> William -- thumbs up! -- we look very much forward for the next
>> version of MixABEL with 'null' heritability estimation procedure
>> separated from the GWAS part. My results suggest that you may be
>> slightly under-estimating the heritability in fmm.
>
>
> Thanks for the feedback Yurii.
>
> bw
>
> Will
> _______________________________________________
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>