susieJAM_alphas: Compute conditional alphas using SuSiE JAM

In USCbiostats/hJAM: Hierarchical Joint Analysis of Marginal Summary Statistics

Compute conditional alphas using SuSiE JAM

Description

The susieJAM_alphas function is to perform the variable selection and compute the selected conditional alpha vector for one intermediate. If only one intermediate in the model, please use susieJAM_alphas instead of susieJAM_A

Usage

susieJAM_alphas(
  marginalA,
  marginalA_se,
  N.Gx,
  eaf.Gy = NULL,
  Geno,
  L.cs = 10,
  min_abs_corr = 0.6,
  max_iter = 100,
  coverage = 0.95,
  estimate_residual_variance = FALSE
)

Arguments

marginalA	the marginal effects of SNPs on one exposure (Gx).
marginalA_se	the standard error of the marginal effects of SNPs on one outcome (Gx).
N.Gx	the sample size of the Gx. It can be a scalar.
eaf.Gy	The vector of the minor allele frequency or effect allele frequency in the GWAS.
Geno	the reference panel (Geno), such as 1000 Genome. The reference data has to be centered.
L.cs	A susie input parameter. Number of components (nonzero elements) in the SuSiE regression model. If L.cs is larger than the number of covariate (p), L.cs is set to p.
min_abs_corr	A susie input parameter. Minimum of absolute value of correlation allowed in a credible set. The default, 0.5, corresponds to squared correlation of 0.25, which is a commonly used threshold for genotype data in genetics studies.
max_iter	Maximum number of iterations in SuSiE fitting.
coverage	Default as 0.95.The coveralge level of the credible set.
estimate_residual_variance	Default as TRUE. Estimate the residual variance in each iteration of SuSiE fitting.

Author(s)

Lai Jiang

Examples

data(GTEx.PrCa)
include.SNPs = which(GTEx.PrCa.inclusion.indicator[,1]==1)
susieJAM_alphas(marginalA = GTEx.PrCa.marginal.A[include.SNPs, 1],
marginalA_se = GTEx.PrCa.marginal.A.se[include.SNPs, 1], eaf.Gy = GTEx.PrCa.maf.gwas[include.SNPs],
Geno = GTEx.PrCa.Geno[, include.SNPs], N.Gx = 620, L.cs = 10, min_abs_corr = 0.5)

USCbiostats/hJAM documentation built on Jan. 26, 2024, 5:27 p.m.