Inverse-variance weighted GWAS meta-analysis

Performs a fixed-effects inverse-variance weighted (IVW) meta-analysis across multiple cohort parquet files. Processes each chromosome via meta_analyze_chromosome(), applies a minor allele count (MAC) filter, and returns heterogeneity statistics (Q, I2) alongside the weighted effect estimates.

Usage

meta_analyze_ivw(
  parquet_files,
  trait_type,
  chromosomes = 1:22,
  min_mac = 100,
  se_col = SE,
  beta_col = B,
  eaf_col = EAF,
  chr_col = CHR,
  pos_col = POS_38,
  rsid_col = RSID,
  ea_col = EffectAllele,
  oa_col = OtherAllele,
  n_col = N,
  effective_n_col = EffectiveN,
  case_n_col = CaseN,
  control_n_col = ControlN
)

Arguments

parquet_files

Character vector of paths to prepared parquet files (output of prep_gwas()).

trait_type

"binary" or "quantitative" (required).

chromosomes

Integer vector of chromosomes to process. Default 1:22.

min_mac

Minimum minor allele count. Variants below this threshold are excluded. Default 100.

se_col

Bare column name for SE. Default SE.

beta_col

Bare column name for effect size. Default B.

eaf_col

Bare column name for EAF. Default EAF.

chr_col

Bare column name for chromosome. Default CHR.

pos_col

Bare column name for position. Default POS_38.

rsid_col

Bare column name for rsid. Default RSID.

ea_col

Bare column name for effect allele. Default EffectAllele.

oa_col

Bare column name for other allele. Default OtherAllele.

n_col

Bare column name for total N. Default N.

effective_n_col

Bare column name for effective N. Default EffectiveN.

case_n_col

Bare column name for case N. Default CaseN.

control_n_col

Bare column name for control N. Default ControlN.

Value

A tibble::tibble() with columns: CHR, POS_38, RSID, EffectAllele, OtherAllele, B, SE, p_value, z_score, EAF, n_contributions, N, EffectiveN, Q, Q_df, Q_pval, I2. Binary traits additionally include CaseN and ControlN.

Details

IVW meta-analysis across cohorts

Examples

if (FALSE) { # \dontrun{
meta <- meta_analyze_ivw(
  parquet_files = c("study1.parquet", "study2.parquet"),
  trait_type    = "binary",
  chromosomes   = 1:22
)
} # }