pseudoAicBic: Function to obtain pseudo-AIC and pseudo-BIC estimates given...

Description Usage Arguments Details

View source: R/pseudoAicBic.R

Description

Function to obtain pseudo-AIC and pseudo-BIC estimates given summary statistics and a reference panel

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
pseudoAicBic(
  penalizedBetas,
  betas,
  ses,
  N,
  refPanel,
  sigSqReg = 0.2,
  sseReg = 0.2,
  sigSqInd = NULL,
  allele1 = NULL,
  allele2 = NULL,
  standardized = TRUE,
  extract = NULL,
  ldBlocks = NULL
)

Arguments

penalizedBetas

Polygenic risk score effect size estimates. These are the estimates for which we want to approximate AIC and BIC. May be a matrix.

betas

Univariate effect size estimates.

ses

Standard errors of the univariate effect size estimates.

N

Sample sizes corresponding to the univariate effect size estimates. Can be a constant or a vector.

refPanel

Stem of PLINK binary file

sigSqReg

The matrix regularization parameter for the estimation of the residual variance. Default is .2.

sseReg

The matrix regularization parameter for the estimation of the SSE. Default is .2.

sigSqInd

Index of SNPs to be used for estimation of residual variance. If null, default is to pick the n/5 SNPs with largest univariate effect sizes, where n is the sample size of the reference panel.

allele1

Vector of effect alleles for the betas and penalizedBetas. Corresponds to the fifth column of a PLINK .bim file.

allele2

Vector of reference alleles for the betas and penalizedBetas. Corresponds to the sixth column of a PLINK .bim file.

standardized

Set to true if the coefficient estimates for penalizedBetas are standardized. Note that elastSum and tlpSum output standardized estimates.

extract

Vector of the indices of SNPs to keep. If null, will keep all SNPs.

ldBlocks

Location of file specifying independent LD Blocks to be used. File should be in the BED file format. If null, estimation is done by chromosome.

Details

Function requires univariate effect size estimates, standard errors, and sample sizes, candidate polygenic risk score estimates, and a reference panel.


jpattee/penRegSum documentation built on Jan. 23, 2021, 12:59 a.m.