TwoSampleMR.Rmd

---
title: "Two-sample MR - Diabetes Dementia"
author: "Mingzhou_Fu"
date: "1/15/2021"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

```{r message=FALSE}
# library(devtools)
# install_github("MRCIEU/TwoSampleMR")
library(TwoSampleMR)
```

```{r}
# List available GWASs
ao = available_outcomes()
# Get instruments
exposure_dat_1 = extract_instruments("ieu-a-23",   # Mahajan,DIAGRAM,2014
                                     p1 = 1,   # significance threshold
                                     clump = FALSE) 
exposure_dat_2 = extract_instruments("ieu-a-24", # Morris,DIAGRAMplusMetabochip,2012
                                     p1 = 1,   # significance threshold
                                     clump = FALSE) 
exposure_dat_3 = extract_instruments("ebi-a-GCST006867", # Xue, 2018
                                     p1 = 1,   # significance threshold
                                     clump = FALSE)
# This GWAS is not working, we need to extract summary statistics from the original file
# However, this is not working because 
library(R.utils)
library(readr)
Xue_T2DM = gunzip("30054458-GCST006867-EFO_0001360-build37.f.tsv.gz", "30054458-GCST006867-EFO_0001360-build37.f.tsv")
Xue_T2DM = read_tsv("30054458-GCST006867-EFO_0001360-build37.f.tsv")
Xue_T2DM = as.data.frame(Xue_T2DM)
# Then we modify the df to be similar to the extracted ones
exposure_xue = 
  Xue_T2DM %>% 
  rename(beta.exposure = beta,
         pval.exposure = p_value,
         se.exposure = standard_error,
         chr.exposure = chromosome,
         samplesize.exposure = n,
         pos.exposure = base_pair_location,
         effect_allele.exposure = effect_allele,
         other_allele.exposure = other_allele,
         eaf.exposure = effect_allele_frequency,
         SNP = variant_id) %>% 
  mutate(id.exposure = "ebi-a-GCST006867") %>% 
  mutate(exposure = "Type 2 diabetes || id:ebi-a-GCST006867") %>% 
  mutate(mr_keep.exposure = TRUE) %>% 
  mutate(pval_origin.exposure = "reported") %>% 
  mutate(data_source.exposure = "igd") %>% 
  select(beta.exposure, pval.exposure, se.exposure, chr.exposure, samplesize.exposure, pos.exposure, id.exposure,
         SNP, effect_allele.exposure, other_allele.exposure, eaf.exposure, exposure, mr_keep.exposure, pval_origin.exposure, data_source.exposure)

# Get effects of instruments on outcome
outcome_dat = extract_outcome_data(snps = exposure_dat_2$SNP, outcomes = "ieu-b-2") # Kunkle,2019
outcome_dat_xue = extract_outcome_data(snps = exposure_xue$SNP, outcomes = "ieu-b-2") # Kunkle,2019
# Harmonise the exposure and outcome data
dat = harmonise_data(exposure_dat_2, outcome_dat)
dat_2 = harmonise_data(exposure_xue, outcome_dat_xue)
# Perform MR
res = mr(dat)
res_2 = mr(dat_2)
View(res)


```