Cullis_H2: Calculate Generalized Heritability from lme4 Model

In etnite/bwardr: Brian Ward personal R package

Calculate Generalized Heritability from lme4 Model

Description

Calculate Generalized Heritability from lme4 Model

Usage

Cullis_H2(model, geno_label = "GENO")

Arguments

model	A lme4 model object
geno_label	A string denoting the label of the random genotypic effect in the supplied lme4 model object

Details

This function calculates generalized heritability using the method of Cullis et al., 2006 (https://doi.org/10.1198/108571106X154443). Specifically, their formula is H2 = 1 - (vblup / (2 * var_g)). Where the generalized heritability (H2) is a function of the reliability of the BLUPs (vblup - the average standard error of differences between BLUPs squared), and the genotypic variance (var_g). This method can be used in unbalanced applications where the traditional entry-mean heritability calculation will give biased estimates. The method of doing this using lme4 is detailed by Ben Bolker at https://stackoverflow.com/questions/38697477/mean-variance-of-a-difference-of-blues-or-blups-in-lme4. This method yields values that are slightly different (I have observed up to 0.75%) from ASReml-R's results. Another solution I came across at https://shantel-martinez.github.io/resources.html seems to produce results that are more divergent from ASReml-R's.

Value

A list containing the following elements:

• avsed The average standard error of differences between adjusted means estimates

• H2 The generalized heritability estimate

etnite/bwardr documentation built on Jan. 6, 2023, 7:12 a.m.