ibdhap.compare.loci: ibdhap compare.loci

Description Usage Arguments Value Author(s) Examples

Description

Compares inferred ibd state with simulated "true" states. Calculates the proportion of markers called in the correct state, false positives (i.e. inferring ibd shared when none is shared), false negatives (i.e. inferring no ibd shared when ibd sharing is present) and the proportion of no calls.

Usage

1
ibdhap.compare.loci(calls, true, data.type)

Arguments

calls

The data.frame created from running ibdhap.make.calls on ibd_haplo output.

true

The data frame of same dimension as calls, but with true (probably simulated) ibd states.

data.type

"h" : haplotypic data "g" : genotypic data (or hap data ran as genotypic) "r" : reduced data (output from running ibdhap.reduce.states and then ibdhap.make.states)

Value

Returns a list consisting of three matrices called "all", "ibd", "nonibd" and "categories".

The matrix for "all" contain a row for each of the following quantities, for all the loci:

Number of sites
Called Correctly
Called IBD Incorrectly
Called noIBD Incorrectly

The matrix for "ibd" contains a row for each of the following quantities, based on just the loci that are truly in an IBD state:

Number of sites
Called Correctly
Called as wrong IBD
Called as no IBD
No-call

The matrix for "nonibd" contains a row for each of the follwing quantities, based on just the loci that are truly in an non-ibd state:

Number of sites
Called Correctly
Called as IBD
No-call

Finally, the matrix for "categories" tabulates the percentage of the loci that are truly in, and called to be in, each IBD state.

Author(s)

Fiona Grimson

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
## this example is taken from the package vignette.
##See vignette(IBDhaploRtools_tutorial)

data(qibd_phased)
data(ids_phased)
data(trueibd_phased)

phased.gold <- ibdhap.make.calls( qibd.file  = qibd_phased,
                   ids.file= ids_phased, cutoff = 0.8)

ibdhap.compare.loci(phased.gold, trueibd_phased, "h")

IBDhaploRtools documentation built on May 2, 2019, 6:48 p.m.