plink_pca: PCA with PLINK

In imbs-hl/imbs: Collection of functions used at the IMBS

Description

PCA with PLINK

Usage

plink_pca(bfile, output.prefix, num.evec, outlier.removal = FALSE,
  outlier.sigma, num.outlier.evec, num.outlier.iter, ld.pruning.params,
  bed.file = NULL, bim.file = NULL, fam.file = NULL,
  exec = "plink2", tmp.dir = NULL, num.threads, memory)

Arguments

bfile	[string] The basename of the binary PLINK files.
output.prefix	[string] The basename of the new binary PLINK files.
num.evec	[int] Number of principal componentns to calculate.
outlier.removal	[flag] Indicate if outliers shall be removed before final PCA computation.
outlier.sigma	[int] How many standard deviations from the mean of a single PC is not an outlier?
num.outlier.evec	[int] How many PCs shall be used for outlier detection?
num.outlier.iter	[int] Number of outlier iterations.
ld.pruning.params	[list] List with function arguments passed to plink_ld_pruning.
bed.file	[string] Alternative to bfile interface. Specify bed, bim and fam files individually.
bim.file	[string] Alternative to bfile interface. Specify bed, bim and fam files individually.
fam.file	[string] Alternative to bfile interface. Specify bed, bim and fam files individually.
exec	[string] Path of PLINK executable.
tmp.dir	[string] Path where to save temporary files. If not set by user, defaults to tempdir().
num.threads	[int] Number of CPUs usable by PLINK. Default is determined by SLURM environment variables and at least 1.
memory	[int] Memory for PLINK in Mb. Default is determined by minimum of SLURM environment variables SLURM_MEM_PER_CPU and num.threads * SLURM_MEM_PER_NODE and at least 5000.

Details

First independent SNPs are extracted using PLINK LD pruning (plink_ld_pruning). If requested, outliers are detected and removed by calculating a PCA, calculating the mean (\hat{μ}_{j}) and standard deviation (\hat{σ}_{j}) of each principal component, and excluding those samples who's PC value v_{j} exceeds \hat{μ}_{j}\pmν\,\hat{σ}_{j} for at least one j, where ν is outlier.sigma.

Value

A list of logs, Eigenvectors and Eigenvalues of both the outlier removal and the final PCA.

imbs-hl/imbs documentation built on Sept. 6, 2019, 11:05 p.m.