af_comparison

af_comparison.rmd

---
title: Comparison of effect allele frequencies in TwoSampleMR
author: Gibran Hemani
date: "`r Sys.time()`"
---


```{r, echo=FALSE}
library(knitr)
opts_chunk$set(warning=FALSE, message=FALSE)
```

## Load libraries

```{r}
library(TwoSampleMR)
library(ieugwasr)
library(tidyverse)
```

## Get allele frequencies for MR analysis

Get harmonised data for exposure and outcome
e.g. BMI vs CHD

```{r}
dat <- 	make_dat(exposures="ieu-a-2", outcomes="ieu-a-7")
glimpse(dat)
```

The `eaf.exposure` and `eaf.outcome` columns are present here. Can plot them

```{r}
plot(eaf.exposure ~ eaf.outcome, data=dat)
```

## Inferring the control sample allele frequencies

Add meta data to get case control sample sizes for CHD - sometimes these are available per-SNP but not in this case

```{r}
dat <- add_metadata(dat)
dat$ncase.outcome
dat$ncontrol.outcome
```

Infer the AF for controls only for CHD. Guess that CHD population prevalence is 0.2, but just made this up

```{r}
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
)
```

Plot again

```{r}
plot(eaf.exposure ~ eaf.outcome.controls, data=dat)
```


## Compare against 1000 genomes super populations

Look up the allele frequencies for these SNPs in 1000 genomes

```{r}
ref <- afl2_rsid(dat$SNP)
```

Compare allele frequencies of exposure, outcome, and reference populations

```{r}
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()
```