af_comparison

af_comparison.r

library(TwoSampleMR)
library(ieugwasr)
library(tidyverse)

# Get harmonised data for exposure and outcome
# e.g. BMI vs CHD
dat <- 	make_dat(exposures="ieu-a-2", outcomes="ieu-a-7")

# infer the AF for controls only for CHD
# first get case control sizes
dat <- add_metadata(dat)
dat$eaf.outcome.controls <- get_population_allele_frequency(
	af = dat$eaf.outcome,
	prop = dat$ncase.outcome / (dat$ncase.outcome + dat$ncontrol.outcome),
	odds_ratio = exp(dat$beta.outcome),
	prevalence = 0.2
)

# This includes the effect allele frequencies
# e.g. we can plot them
plot(dat$eaf.exposure ~ dat$eaf.outcome)

# Look up the allele frequencies for these SNPs in 1000 genomes
ref <- afl2_rsid(dat$SNP)

# Compare allele frequencies of exposure, outcome, and reference populations
dat %>%
	select(SNP, eaf.exposure, eaf.outcome.controls) %>%
	inner_join(., ref, by=c("SNP"="rsid")) %>%
	select(eaf.exposure, eaf.outcome.controls, starts_with("AF.")) %>%
	pairs()