Tutorial 4: Meta analysis

#library(gwasglue2)
library(ieugwasr)
devtools::load_all("../") # this was added just for development

Meta-analysis is a statistical combination of the results from two or more separate studies. In gwasglue2, we use the fixed-effect model which assumes that one true effect size underlies all the studies in the meta-analysis.

We are going to perform meta-analysis for two different studies of cardiac heart disease (chd), in the HMGCR (3-hydroxy-3-methylglutaryl-CoA reductase) gene region.

Firt, we choose the IEU ids for the chd trait.

ids <- c( "ieu-a-7",  "ukb-d-I9_IHD")

Then, we obtain the metadata using ieugwasr::gwasinfo() for each study and create a metadata object.

metadata <- lapply(seq_along(ids), function(i){
  m <- create_metadata(ieugwasr::gwasinfo(ids[i])) 
})

In the code bellow, we create an harmonised dataset object from the summary sets for each study.

#  create dataset 
  hmgcr_chrpos <- "5:74132993-75132993"
  dataset <- lapply(seq_along(ids), function(i){
    # create summarysets
    s <- create_summaryset(ieugwasr::associations(variants = hmgcr_chrpos, id =ids[i]), metadata=metadata[[i]])
  }) %>%
    # create dataset
    create_dataset(., harmonise = TRUE, tolerance = 0.08, action = 1)

Finally, we perform the meta-analysis in

to create a new summary set

meta_chd <- dataset%>%
   meta_analysis(.)