In bogdanlab/RHOGE: Genome-wide genetic correlation between two complex traits using TWAS effect-size estimates

RHOGE is an R package that estimates the genome-wide genetic correlation between two complex traits (diseases) as a function of predicted gene expression effect on trait (\rho_{ge}). Given output from two transcriptome-wide association studies, RHOGE estimates the mediating effect of predicted gene expression and estimates the correlation of effect sizes across traits (diseases). This approach is extended to a bi-directional regression that provides putative causal directions between traits with non-zero \rho_{ge}.

This approach is described in:

Integrating Gene Expression with Summary Association Statistics to Identify Genes Associated with 30 Complex Traits \ Nicholas Mancuso, Huwenbo Shi, Pagé Goddard, Gleb Kichaev, Alexander Gusev, Bogdan Pasaniuc.\ American Journal of Human Genetics. 2017.

Installation

Bioconductor + devtools is the most straightforward way to install RHOGE. To do this open an R terminal and enter

source("http://bioconductor.org/biocLite.R")
biocLite("devtools")    # only if devtools not yet installed
biocLite("bogdanlab/RHOGE")

Example

The following example computes \rho_{ge} between BMI and triglycerides, as well as putative causal directions.

library(RHOGE)

# example BMI TWAS results from FUSION
data(bmi)
head(bmi)

# example triglyceride TWAS results from FUSION
data(tg)
head(tg)

# Estimate rho_ge genome-wide for BMI and Triglyerides and approximate sample sizes
ge_cor_res <- rhoge.gw(bmi, tg, 14000, 91000)
head(ge_cor_res)

# Perform bi-directional regression to estimate putative causal directions
bidir_res <- rhoge.bd(bmi, tg, 14000, 91000, p1 = 0.05 / nrow(bmi), p2 = 0.05 / nrow(tg))
head(bidir_res)

Notes

Currently, only FUSION style output is supported.

RHOGE comes installed with estimates of approximately independent LD blocks for European, Asian, and African ancestries. Performance should improve if you have in-sample estimates of LD blocks. The only requirement is that regions are stored as a data.frame-like object with 3 columns ('CHR', 'START', 'STOP'). For example,

library(RHOGE)
data("grch37.eur.loci")
head(grch37.eur.loci)

bogdanlab/RHOGE documentation built on May 25, 2019, 3:24 p.m.