From listmaster animalgenome.org Wed Nov 4 11:20:49 2020
Return-Path: <listmasteranimalgenome.org>
From: Hong Lee <Hong.Leeunisa.edu.au>
Postmaster: submission approved by list moderator
To: Members of AnGenMap <angenmapanimalgenome.org>
Subject: MTG2 (CORE GREML)
Date: Wed, 04 Nov 2020 11:20:49 -0600
Dear AnGenMap members
I would like to introduce a function in MTG2 that can estimate correlations
between random effects between which covariance structure is not pre-defined,
i.e. CORE (G)REML.
Classical linear mixed models assume that such correlations are negligible
(e.g. the correlation between g and e in a model, y = g + e). This is usually
ok, but if two random effects are different parts of genomic regions (e.g.
regulatory vs. intergenic or rare vs. common variants), this assumption
becomes invalid. This also applies to multi-omics data analysis (e.g. a
linear mixed model fitting genomic and transcriptomic effects simultaneously).
Unless genome-transcriptome correlation is correctly modelled, the analysis
can provide biased estimates of the variance components.
Please see section 15 in MTG2 manual (and binary file and example files) that
can be download from https://sites.google.com/...ite/honglee0707/mtg2
The relevant publication is
- CORE GREML for estimating covariance between random effects in linear
mixed models for complex trait analyses.
Nature Communications. Nature Communications 11: 4208 (2020).
(https://www.nature.com/...s/s41467-020-18085-5)
In addition, MTG2 version 2.17 has been optimised for the computing speed of
multivariate linear mixed models (REML) that is > 10 times faster than earlier
versions when fitting many levels of covariates.
Any question or feedback would be appreciated (hong.leeunisa.edu.au)
Best wishes
Hong
|
