mJAM_build_CS: Construct mJAM credible set based for selected index SNP
In USCbiostats/hJAM: Hierarchical Joint Analysis of Marginal Summary Statistics

mJAM_build_CS

R Documentation

Construct mJAM credible set based for selected index SNP

Description

Construct mJAM credible set based for selected index SNP

Usage

mJAM_build_CS(
  X_id,
  prev_X_list = NULL,
  All_id,
  PrCS_weights = "Pr(M_C)",
  coverage = 0.95,
  GItGI_curr,
  GIty_curr,
  yty_curr,
  yty_med,
  N_GWAS,
  rare_SNPs = NULL,
  Pr_Med_cut = 0.1,
  use_robust_var_est = FALSE
)

Arguments

`X_id`	A character specifying the ID of the index SNP; should be found in 'All_id'.
`prev_X_list`	A list of character vector of the ID(s) of previously selected index SNP(s).
`All_id`	A list of character vector of the ID(s) of all SNP(s) remaining in the analysis, including all previously selected SNP(s) and the current index SNP.
`PrCS_weights`	An option to specify what weights to apply on Pr(Med). Default is "Pr(M_C)".
`coverage`	A number between 0 and 1 specifying the “coverage” of the estimated confidence sets.
`GItGI_curr`	A list of GItGI statistics at the current stage (after pruning out SNPs correlated with previously selected index SNPs).
`GIty_curr`	A list of GIty estimates of all remaining SNPs at the current stage (after pruning out SNPs correlated with previously selected index SNPs).
`yty_curr`	A list of yty estimates of all remaining SNPs at the current stage (after pruning out SNPs correlated with previously selected index SNPs).
`yty_med`	A list of median yty across all SNPs.
`N_GWAS`	A vector of sample sizes in all original GWAS studies.
`rare_SNPs`	A numeric vector of ID(s) for rare SNP(s) which we do not apply weighting. Instead, we use the individual estimate of yty for these SNPs for robustness.
`Pr_Med_cut`	The cutoff for Pr(Mediation); SNPs with Pr(Mediation) smaller than this cutoff will be assigned a Pr(CS) = 0 and thus not included in the credible set for the current index
`use_robust_var_est`	whether to use linear combination of median yty and individual yty.

Value

A table with the following columns:

CS_SNP: SNP name.
Post_Model_Prob: The posterior Pr(Model) of this SNP on its absolute scale.
Post_Model_Prob_Ratio: The posterior Pr(Model) of this SNP divided by the posterior Pr(Model) of index SNP. It should be <= 1.
Post_Model_Prob_Ratio2: If ‘Post_Model_Prob_Ratio' is greater than 1, set 'Post_Model_Prob_Ratio2' to 1. Otherwise, it’s the same as 'Post_Model_Prob_Ratio'.
Med_Effect_Size: The posterior mediation effect size.
Post_Med_Prob: The posterior Pr(Mediation) of this SNP.
Post_Med_Prob2: If ‘Post_Med_Prob' is less than 'Pr_Med_cut', set 'Post_Med_Prob2' to 0. Otherwise, it’s the same as 'Post_Med_Prob'.
SD_Post_CS_Prob: Standardized Pr(CS) where Pr(CS) = Pr(Model)*Pr(Mediation)
CumSum_Porb: The cumulative 'SD_Post_CS_Prob'. Note that the table is ordered by descending 'SD_Post_CS_Prob'.
EmpiricalCut: The empirical coverage of this CS (should be >= requested 'coverage').
CS_in: A logical variable indicating whether this CS_SNP is included in this CS or not.
index_SNP: The name of the index SNP.