Using linear mixed models for genome-wide interaction analyses

Question

[kaf19]

Hello,

I would like to test SNP*sex interaction terms for association with a quantitative trait. My sample includes related individuals so I would like to use a linear mixed model. My SNPs include both directly genotyped and imputed SNPs in binary PLINK and binary GEN format respectively. Does anyone know of any appropriate software to use for this (the less file conversion required the better!)?

Software such as GEMMA, lrgpr, and Limix (iSet) look like they might do this - does anyone disagree?

Many thanks in advance,

Katherine

[Ahill]

Perhaps EPACTS / EMMAX ? EPACTS link provides instructions on file formats needed.

[Kevin Blighe]

I recently developed a system for analysing trios GWAS data in R with quantitative traits and interaction terms. The input data is exported in non-binary format from PLINK and then pseudocontrols are created. This is essentially doing a variation of PBAT and FBAT.

Based on the information you've provided, I''m not sure if it's directly applicable (or even desirable), but you can take a look: GenomewideTriosCLogit (GitHub).