FEMA-Long is available for use here: github.com/cmig-researc...

All of these features are implemented in a computationally fast manner, thereby allowing scalability to very large datasets as well as large number of outcome variables like voxel-wise or vertex-wise analyses.

FEMA-Long can perform longitudinal GWAS with SNP*time non-linear interaction to discover SNPs showing time-varying effects. The top part of the Miami plots show SNPs having time-dependent effect compared to longitudinal GWAS (bottom part). Last panel shows the effect of a few selected SNPs over time

FEMA-Long can model unstructured covariance such as time-varying heritability and genetic correlations which are super critical for longitudinal datasets. Here, using the MoBa dataset, we show time-varying random effects for length, weight, and BMI in the first year of life.

Massive thank you to my amazing co-authors including @nadineparker.bsky.social @neuromusicnerd.bsky.social @jaholkowskipiotr.bsky.social @cmakowski.bsky.social @nichols.bsky.social and others who are not on BlueSky. Wouldn't have been possible without their help, support, kindness, and patience :)

 “people won’t take us seriously if we speak plainly about what we’re doing, so we have to smoke and mirror some fake math at them to get them to show us respect”.