Relation to LIMIX

CellRegMap’s linear mixed model (LMM) uses the FaST-LMM (Factored Spectrally Transformed Linear Mixed Model) implementation described here and used within the LIMIX framework.

Linear Mixed Model implementation using LIMIX

LIMIX is described in this preprint and documentation can be found here. While there are several fast implementations to map eQTLs using correlation or linear regression-based approaches (e.g., tensorQTL, matrix eQTL), to our knowledge LIMIX is the fastest software out there for eQTL association analyses using linear mixed models (which allow to better model population stratification and cryptic relatedness, which are prevalent in human genetic data - see for example thread here).

eQTL mapping using LIMIX

Specifically, for a LIMIX wrapper to map eQTLs, see the limix QTL pipeline developed in our lab, which we most recently used in our Genome Biology Publication.

The pipeline provides an easy wrapper to map eQTLs using various methods, automatically re-ordering, subselecting and expanding files to match with each other. See the wiki pages for installation and input files, and an example snakemake for standard eQTL mapping.

Coming Soon

With Marc Jan Bonder, we are in the process of implementing CellRegMap runners compatible with this pipeline.

Also see our sister project scDALI, a model for modelling allelic imbalance in single cells.