Skip to content
R wrapper to python module limmbo for covariance matrix estimation in multivariate linear mixed models in genetics studies
R
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
R
man
tests
.Rbuildignore
.gitignore
.travis.yml
DESCRIPTION
LICENSE
NAMESPACE
NEWS.md
README.Rmd
README.md
limmbo2.Rproj

README.md

Travis-CI Build Status

limmbo2

The goal of limmbo2 is to estimate Vg and Ve covariance matrices in the multivariate LMM. Note that it uses the python module limmbo to do this. The preprint article that describes the limmbo python module's methods is here: https://www.biorxiv.org/content/early/2018/01/30/255497

Installation

First, be sure that you have python modules limix and limmbo installed:

conda install -c conda-forge limix
pip install limmbo

Once those two modules are successfully installed, you can proceed to install the limmbo2 R package:

# install.packages("devtools")
devtools::install_github("fboehm/limmbo2")

Example

This is a basic example which shows you how to solve a common problem:

library(limmbo2)
pheno <- matrix(data = runif(300), nrow = 100, ncol = 3)
kinship <- diag(100)
t(chol(kinship)) -> chol_kin

prep_data(pheno, kinship) -> input_data

make_limmbo(input_data, TRUE, 10, 2) -> l_out
bs_covar_est(l_out, 1, 1) -> bs_out
bs_out2 <- lapply(FUN = convert_for_bscombine, X = bs_out)

combine_bs(bs_out2, l_out) -> fits
(fits$Cn_fit -> Ve)
#>           [,1]      [,2]      [,3]
#> [1,] 0.1696938 0.1342838 0.1126630
#> [2,] 0.1342838 0.1813464 0.1230871
#> [3,] 0.1126630 0.1230871 0.1421201
(fits$Cg_fit -> Vg)
#>           [,1]      [,2]      [,3]
#> [1,] 0.1696938 0.1342838 0.1126630
#> [2,] 0.1342838 0.1813464 0.1230871
#> [3,] 0.1126630 0.1230871 0.1421201
You can’t perform that action at this time.