Skip to contents

Enrichment Analysis

Source: R/enrich.R
enrich.Rd

Tests for annotation enrichment using stratified LDSC results.

Usage

enrich(
  s_covstruc,
  model = "",
  params = NULL,
  fix = "regressions",
  std.lv = FALSE,
  rm_flank = TRUE,
  tau = FALSE,
  base = TRUE,
  toler = NULL,
  fixparam = NULL
)

Arguments

s_covstruc

Stratified LDSC result (list with S_baseline, S_annot, V_annot, annotation_names, m_annot, m_total)

model

lavaan-style model syntax (default "" = use basic proportional enrichment test)

params

Character vector of parameter names to test (default NULL = all free params)

fix

Which parameters to fix: "regressions" (default), "loadings", or "none"

std.lv

Standardize latent variables (default FALSE)

rm_flank

Remove flanking regions (default TRUE; implemented in Rust s_ldsc engine)

tau

Use tau parameterization (default FALSE; not yet used)

base

Include baseline (default TRUE)

toler

Gradient tolerance for optimizer (default NULL = auto)

fixparam

Named list of parameters to fix at specific values (default NULL)

Value

A data frame with enrichment results

Details

When model is provided, fits a SEM per annotation using usermodel and tests for parameter differences vs baseline. When model is empty, uses the fast Rust proportional enrichment test.

Examples

if (FALSE) { # \dontrun{
# Fast proportional enrichment test (default, model = "").
s_covstruc <- s_ldsc(
  traits = c("T1.sumstats.gz", "T2.sumstats.gz"),
  ld = "baseline_LD/", wld = "weights/", frq = "frq/"
)
result <- enrich(s_covstruc)
head(result)
} # }