coindex: Calculate Coincidence Index
In etnite/bwardr: Brian Ward personal R package

coindex

R Documentation

Calculate Coincidence Index

Description

Calculate Coincidence Index

Usage

coindex(obs, pred, si, best = "high", naive = FALSE)

Arguments

`obs`	A named vector of observed phenotypic data
`pred`	A named vector of predicted phenotypic data
`si`	Selection index - value from 0 to 1
`best`	String or number signifying what phenotypic values are considered "best." Must be one of the following: "high" to perform directional selection of genotypes with highest trait values "low" to perform directional selection of genotypes with lowest trait values "mean" to perform stabilizing selection around the mean values of the observed and predicted data "median" to perform stabilizing selection around the median values of the observed and predicted data numeric, in which case it is assumed that the ideal phenotypic value has been supplied
`naive`	Logical - if TRUE then a simple proportion matching coefficient is returned, without correcting for random chance

Details

This function calculates the coincidence index as defined by Hamblin and Zimmerman, 1986. (https://doi.org/10.1002/9781118061015.ch8). In the context of this function, its intended use is to compare direct phenotypic selection with indirect selection based upon genomic data. However, it can be used to compare selections made using any two different methods. The formula for the coincidence index is

CI = (C - R) / (T - R)

where T is the number of observed genotypes selected using the selection index, R is the expected number of genotypes among these correctly selected due to chance, and C is the number of coincident genotypes selected using both methods.

For instance, if we have a population of 100 genotypes and a selection index of 0.1, T = 100 * 0.1 = 10 and R = 10 * 0.1 = 1. The value of C will depend upon how many genotypes are selected based upon both observed and predicted data.

A value of 1 indicates identical selections made using the observed and predicted phenotypic values, while a value of 0 indicates that selections based on predicted values were no better than random chance. Note that CI values can be negative. Small samples sizes will lead to high CI variance.