gibran hemani explodecomputer

Tidyverse notes

Piping:

library(dplyr)

	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

	# on bc3

	# wget https://fluff.bris.ac.uk/fluff/u2/mk19726/Al7MuuePNzS7tMWliJGG6QznF/fev1_adjusted_only_for_sex_imputed.txt
	# wget https://fluff.bris.ac.uk/fluff/u2/mk19726/0ar7Sekl4I3Xj4JyatEf_wznc/fvc_adjusted_only_for_sex_imputed.txt
	# wget https://fluff.bris.ac.uk/fluff/u4/mk19726/pUNC547Sala4I5Y1V_G9Bwzna/obstructive_ratio_spirometry_imputed.txt



	# FEV1:
	# sample size 353315

	library(GwasDataImport)
	library(data.table)
	library(ieugwasr)
	library(dplyr)


	# Download the Howard et al 2019 MDD GWAS from https://datashare.ed.ac.uk/handle/10283/3203
	a <- data.table::fread("https://datashare.ed.ac.uk/bitstream/handle/10283/3203/PGC_UKB_depression_genome-wide.txt?sequence=3&isAllowed=y")

	# It doesn't have chr/pos, so let's look them up on OpenGWAS (a bit slow!)

	---
	title: Survival bias in sex association with dementia
	author: Gibran Hemani
	date: `r Sys.time()`
	---

	```{r}
	library(dplyr)
	library(tidyr)
	library(ggplot2)

	# sample size
	n <- 10000

	# simulate x and y variables - no association
	x <- rnorm(n)
	y <- rnorm(n)

	# selected into the sample is influenced by x and y
	sel <- rbinom(n, 1, plogis(x + y))

	library(dplyr)
	library(org.Hs.eg.db)

	# get KEGG pathways

	xx <- as.list(org.Hs.egPATH2EG)
	keggpathways <- lapply(names(xx), function(x) {
	tibble(
	symbol=select(org.Hs.eg.db, xx[[x]], c("SYMBOL"), "ENTREZID")$SYMBOL,
	pathway=x