pops: Bayesian Polygenic Prioritisation Scoring (Bayesian POPS)

In psoerensen/qgg: Statistical Tools for Quantitative Genetic Analyses

Bayesian Polygenic Prioritisation Scoring (Bayesian POPS)

Description

This function performs Polygenic Prioritisation Scoring (POPS) using Bayesian regression ('bayesC' or 'bayesR') or ridge regression ('rr'). It maps features to sets, performs optional feature selection based on p-value thresholds, and calculates predictive scores for prioritisation.

Usage

pops(
  stat = NULL,
  sets = NULL,
  validate = NULL,
  threshold = NULL,
  method = "bayesC",
  pi = 0.001,
  nit = 5000,
  nburn = 1000,
  updateB = TRUE,
  updateE = TRUE,
  updatePi = TRUE,
  updateG = TRUE
)

Arguments

stat	A numeric vector or matrix of summary statistics (e.g., phenotypic values or effect sizes), where rows represent features (e.g., SNPs) and columns represent traits. Required.
sets	A list of feature sets (e.g., genes or SNP groups) to map to the rows of 'stat'. Required.
validate	An optional validation set. If provided, cross-validation results are returned instead of fitting the model.
threshold	A numeric value specifying a p-value threshold for feature selection. If provided, only features with p-values below this threshold are included in the model.
method	A string specifying the regression method. Options are '"bayesC"' (default), '"bayesR"', or '"rr"' (ridge regression).
pi	A numeric value specifying the proportion of non-zero effects for Bayesian methods. Default is '0.001'.
nit	An integer specifying the number of iterations for Bayesian methods. Default is '5000'.
nburn	An integer specifying the number of burn-in iterations for Bayesian methods. Default is '1000'.
updateB	A logical value indicating whether to update marker effects in Bayesian methods. Default is 'TRUE'.
updateE	A logical value indicating whether to update residual variances in Bayesian methods. Default is 'TRUE'.
updatePi	A logical value indicating whether to update the proportion of non-zero effects in Bayesian methods. Default is 'TRUE'.
updateG	A logical value indicating whether to update the genomic variances in Bayesian methods. Default is 'TRUE'.

Value

A matrix of predicted prioritisation scores ('ypred') for each feature, ordered by their predictive values. If a validation set is provided, cross-validation results are returned instead.

psoerensen/qgg documentation built on March 29, 2025, 6:36 p.m.