Collapse high correlation SNPs — collapseHighCorSNPs • mrlocus

A helper function to collapse sets of highly correlated SNPs within signal clusters. This is recommended to run before flipAllelesAndGather, and before fitBetaColoc.

collapseHighCorSNPs(
  sum_stat,
  ld_mat,
  ld_mat2 = NULL,
  threshold = 0.95,
  score = NULL,
  plot = TRUE,
  snp_id = NULL
)

Arguments

sum_stat: list of summary statistic tables, which is a list over signal clusters. Each element of the list should be a data.frame describing the eQTL and GWAS summary statistics. The only column in sum_stat that is used by the function is score (optional)
ld_mat: list of LD matrices across signal clusters
ld_mat2: optional second list of LD matrices (for different populations). it will be returned alongside the first ld_mat, which is used for the collapsing. The second list of LD matrices is just subset to the same set of SNPs as the first
threshold: threshold on absolute value of correlation for collapsing, e.g. will collapse if SNPs are more correlated (or anti-correlated) than this amount
score: name of a column of sum_stat data.frames with a score, such that collapsing will choose the highest score SNP per collapsed cluster. Otherwise, if set to NULL, the first SNP will be used
plot: logical, draw a before/after grid of plots
snp_id: name of SNP id column in sum_stat, if specified will output a column collapsed that lists which SNP ids are represented in the output (i.e. which highly correlated SNPs were collapsed).

Value

list with subset ld_mat and sum_stat

lists (and ld_mat2 if provided)