pseudovalidation: Performs 'pseudovalidation' to select the best lambda value...

In SeojinHwang/scadsum: nonconvex penalty with summary statistics and a reference panel

Performs ‘pseudovalidation’ to select the best \lambda value in lassosum

Description

Performs ‘pseudovalidation’ to select the best \lambda value in lassosum

Usage

pseudovalidation(
  bfile,
  beta,
  cor,
  sd = NULL,
  keep = NULL,
  extract = NULL,
  exclude = NULL,
  remove = NULL,
  chr = NULL,
  cluster = NULL,
  ...
)

Arguments

bfile	A plink bfile stem
beta	The matrix of estimated \betas
cor	The vector of correlations (r)
sd	The standard deviation of the SNPs
keep	samples to keep
extract	SNPs to extract
exclude	SNPs to exclude
remove	samples to remove
chr	a vector of chromosomes
cluster	A cluster object from the parallel package for parallel computing

Details

A function to calculate

f(\lambda)=\beta'r/\sqrt{\beta'X'X\beta}

where X is the standardized genotype matrix divided by \sqrt n, and r is a vector of (shrunken) correlations.

Value

the results of the pseudovalidation, i.e. f(\lambda)

Note

• Missing genotypes are interpreted as having the homozygous A2 alleles in the PLINK files (same as the --fill-missing-a2 option in PLINK).

• The number of rows in beta and the length of cor should be the same as the number of SNPs in the bfile after extract/exclude/chr.

SeojinHwang/scadsum documentation built on June 30, 2023, 10:52 p.m.