phelex: Estimate misclassification in phenotype

In afrahshafquat/phelex: Estimate misclassification probabilities for GWAS Phenotypes

Description

Estimates misclassification probabilities in observed GWAS phenotype y given genotypes dataset x. The method follows the PheLEx algorithm to predict misclassification probabilities using Adaptive Metropolis-Hastings and mixed model defined by Shafquat et al.

Usage

phelex(x, y, A = NULL, iterations = 1e+05, alpha.prior = c(10, 1),
  lambda.prior = c(1, 1), link = "pnorm", beta.prior = "norm",
  beta.prior.params = c(0, 1), beta.initial.vec = NULL,
  sigmaA.initial = 0.001, beta.iterations = 2e+05, beta.warmup = 20000,
  verbose = TRUE, stamp = 10000)

Arguments

x	Genotype matrix with dimensions n x m.
y	Phenotype vector with length n.
A	Genetic similarity matrix with dimensions n x n.
iterations	Total number of iterations
alpha.prior	Beta prior parameters for true positve rate
lambda.prior	Beta prior parameters for false positive rate
link	probit or logistic model (options pnorm and plogis respectively)
beta.prior	'norm'(default): Normal prior or 'unif' Uniform prior on fixed effects
beta.prior.params	if beta.prior is norm, then (mean, sd), if unif then (min, max)
beta.initial.vec	Initial values for fixed effects
sigmaA.initial	Starting value for variance parameter sigmaA where u ~ MVN(0, sigmaA * A).
beta.iterations	Number of iterations to sample fixed effects (should be greater than beta.warmup)
beta.warmup	Burn-in/warmup for initial sampling for fixed effects
verbose	Default TRUE. Prints progress information
stamp	Iteration breakpoint to print time

Value

List containing

• betas: Matrix of estimated effect sizes for each SNP (SNPs[rows] x iterations[columns]).

• parameters: Matrix with estimated parameter values[rows] across iterations[columns]. Order is c(sigmaA, alpha, lambda) where alpha = true positive rate, lambda = false positive rate, sigmaA = variance parameter

• misclassified.cases: Matrix of alpha vectors where 1s represent false positives and 0s represent true positives as inferred at each iterations

• misclassified.controls: Matrix of lambda vectors where 1s represent false negatives and 0s represent true negatives as inferred at each iterations

• posterior: Vector of posterior probability across iterations

• accept: Vector of 1/0 values across iterations; 1 indicates proposal was accepted at iteration;0 o/w

afrahshafquat/phelex documentation built on Feb. 5, 2020, 7:44 p.m.