ibs.stats: function to calculate the identity-by-state stats of a group...

In chopsticks: The 'snp.matrix' and 'X.snp.matrix' Classes

Description

Given a snp.matrix-class or a X.snp.matrix-class object with $N$ samples, calculates some statistics about the relatedness of every pair of samples within.

Usage

ibs.stats(x)

Arguments

x	a snp.matrix-class or a X.snp.matrix-class object containing $N$ samples

Details

No-calls are excluded from consideration here.

Value

A data.frame containing $N (N-1)/2$ rows, where the row names are the sample name pairs separated by a comma, and the columns are:

Count	count of identical calls, exclusing no-calls
Fraction	fraction of identical calls comparied to actual calls being made in both samples

Warning

In some applications, it may be preferable to subset a (random) selection of SNPs first - the calculation time increases as $N (N-1) M /2$ . Typically for N = 800 samples and M = 3000 SNPs, the calculation time is about 1 minute. A full GWA scan could take hours, and quite unnecessary for simple applications such as checking for duplicate or related samples.

Note

This is mostly written to find mislabelled and/or duplicate samples.

Illumina indexes their SNPs in alphabetical order so the mitochondria SNPs comes first - for most purpose it is undesirable to use these SNPs for IBS purposes.

TODO: Worst-case S4 subsetting seems to make 2 copies of a large object, so one might want to subset before rbind(), etc; a future version of this routine may contain a built-in subsetting facility to work around that limitation.

Author(s)

Hin-Tak Leung [e-mail]

Examples

data(testdata)
result <- ibs.stats(Autosomes[11:20,])
summary(result)

chopsticks documentation built on Nov. 8, 2020, 7:51 p.m.