qtl2pleio on CRAN

January 10, 2020

CRAN now hosts the packages qtl2pleio and gemma2. qtl2pleio performs a d-variate, d-QTL scan over a select genomic region. gemma2 is used by qtl2pleio for the inference of multivariate variance components.

They can be installed with:

install.packages("qtl2pleio")

The statistical model that qtl2pleio fits for each d-tuple of markers (or pseudomarkers) is

$vec(Y) = Xvec(B) + vec(G) + vec(E)$ where $$Y$$ is a n by d matrix of d traits (for each of n subjects), X is a dn by df block-diagonal matrix of founder allele probabilities, B is a f by d matrix of allele effects for each of d traits, G is a n by d matrix of polygenic random effects, and E is a n by d matrix of random errors. We assume that

$G \sim N(0, K, V_g)$

and

$E \sim N(0, I, V_e)$

We further assume that $$G$$ and $$E$$ are independent. $$K$$ is an estimated kinship matrix.

In a recent G3 article, we developed our test of pleiotropy vs. separate QTL and applied it to two behavioral traits measured on a cohort of Diversity Outbred mice.

qtl2pleio “plays nicely” with qtl2 and other R packages from the qtl2 ecosystem by using the same structures for genotype data, phenotype data, and covariates.

We’re currently investigating our pleiotropy test in the context of expression trait hotspots.