imputeSnpMatrix: Missing values handling for SnpMatrix object

In GeneGeneInteR: Tools for Testing Gene-Gene Interaction at the Gene Level

Description

imputeSnpMatrix is a generic wrapper of snp.imputation and impute.snps functions from snpStats package. This function mimics a Leave-One-Out process where missing SNPs are imputed for an individual based on a model trained on all other individuals.

Usage

imputeSnpMatrix(snpX, genes.info, on.rem = c("SNP", "ind", "none"),
  quiet = FALSE)

Arguments

snpX	snpMatrix object of which SNPs are to be removed
genes.info	A data.frame with four columns named Genenames, SNPnames, Position and Chromosome. Each row describes a SNP and missing values are not allowed.
on.rem	(optional) a character string matching one of the following items: SNP, ind, none. Describes the action taken in case of remaining missing values. See details
quiet	(optional) a boolean describing wether or not progress bar should be displayed.

Details

For the ith row in the snpX argument (i.e. the ith individual in the dataset), the following steps are performed:

• missing SNPs are detected for individual i

• rules are imputed for every missing SNPs using the whole dataset where individual i is removed

• SNPs are imputed for individual i

Although, such a process allows the imputation of a large number of missing SNPs, some missing values may remains. In that case, the user can specify the action to be done thanks to the om.rem arguments:

• om.rem="none": leave the dataset as it is,

• om.rem="SNP": remove all SNPs with remaining missing values,

• om.rem="ind": remove all individuals with remaining missing values.

Removing all SNPs is often more parsimonious than removing individuals and allows to get a dataset without any missing values with minimum information-loss.

Value

a list object with two named elements:

snpX	a SnpMatrix object corresponding to input matrix with imputed values.
genes.info	a data frame object corresponding to an updated version of input genes.info in case SNP had to be removed.

A warning is printed when SNPs or individuals are removed.

See Also

GGI snpMatrixScour

Examples

## Not run: 
ped <- system.file("extdata/example.ped", package="GeneGeneInteR")
info <- system.file("extdata/example.info", package="GeneGeneInteR")
posi <- system.file("extdata/example.txt", package="GeneGeneInteR")
data.imported <- importFile(file=ped, snps=info, pos=posi, pos.sep="\t")

## Example 1 without SNPs filtering
## In this example, 9 SNPs were removed due to remaining missing values.
imputed.snps.1 <- imputeSnpMatrix(data.imported$snpX, genes.info = data.imported$genes.info)


## Example 2 with SNPs filtering priori to the imputation
## Filtering of the data: SNPs with MAF < 0.05 or p.value for HWE < 1e-3 or SNPs with
## call.rate < 0.9 are removed. 
data.scour <- snpMatrixScour(snpX=dta$snpX,genes.info=dta$genes.info,min.maf=0.05,
                              min.eq=1e-3,call.rate=0)


## End(Not run)

## Equivalent loading of data for example 2
## Imputation of the missing genotypes
load(system.file("extdata/dataScour.Rdata", package="GeneGeneInteR"))

data.imputed <- imputeSnpMatrix(data.scour$snpX, genes.info = data.scour$genes.info)

GeneGeneInteR documentation built on Nov. 8, 2020, 6:28 p.m.