LEA (version 1.4.0)

cross.entropy.estimation: compute the cross-entropy criterion

Description

Calculate the cross-entropy criterion. This is an internal function, automatically called by snmf. The cross-entropy criterion is a value based on the prediction of masked genotypes to evaluate the error of ancestry estimation. The criterion will help to choose the best number of ancestral population (K) and the best run among a set of runs in snmf. A smaller value of cross-entropy means a better run in terms of prediction capacity. The cross.entropy.estimation function displays the cross-entropy criterion estimated on all data and on masked data based on the input file, the masked data file (created by create.dataset, the estimation of the ancestry coefficients Q and the estimation of ancestral genotypic frequencies, G (calculated by snmf). The cross-entropy estimation for all data is always lower than the cross-entropy estimation for masked data. The cross-entropy estimation useful to compare runs is the cross-entropy estimation for masked data. The cross-entropy criterion can also be automatically calculated by the snmf function with the entropy option.

Usage

cross.entropy.estimation (input.file, K, masked.file, Q.file, G.file,  ploidy = 2)

Arguments

input.file

A character string containing a path to the input file without masked genotypes, a genotypic matrix in the geno format.

K

An integer corresponding to the number of ancestral populations.

masked.file

A character string containing a path to the input file with masked genotypes, a genotypic matrix in the geno format. This file can be generated with the function, create.dataset). By default, the name of the masked data file is the same name as the input file with a _I.geno extension.

Q.file

A character string containing a path to the input ancestry coefficient matrix Q. By default, the name of this file is the same name as the input file with a K.Q extension.

G.file

A character string containing a path to the input ancestral genotype frequency matrix G. By default, the name of this file is the same name as the input file with a K.G extension (input_file.K.G).

ploidy

1 if haploid, 2 if diploid, n if n-ploid.

Value

cross.entropy.estimation returns a list containing the following components:

masked.ce

The value of the cross-entropy criterion of the masked genotypes.

all.ce

The value of the cross-entropy criterion of all the genotypes.

References

Frichot E, Mathieu F, Trouillon T, Bouchard G, Francois O. (2014). Fast and Efficient Estimation of Individual Ancestry Coefficients. Genetics, 194(4) : 973--983.

See Also

geno create.dataset snmf

Examples

# Creation of tuto.geno
# A file containing 400 SNPs for 50 individuals.
data("tutorial")
write.geno(tutorial.R,"genotypes.geno")

# The following command are equivalent with 
# project = snmf("genotypes.geno", entropy = TRUE, K = 3)
# cross.entropy(project)

# Creation      of the masked data file
# Create file:  "genotypes_I.geno"
output = create.dataset("genotypes.geno")

# run of snmf with genotypes_I.geno and K = 3
project = snmf("genotypes_I.geno", K = 3, project = "new")

# calculate the cross-entropy
res = cross.entropy.estimation("genotypes.geno", K = 3, "genotypes_I.geno",
    "./genotypes_I.snmf/K3/run1/genotypes_I_r1.3.Q", 
    "./genotypes_I.snmf/K3/run1/genotypes_I_r1.3.G")

# get the result
res$masked.ce
res$all.ce

#remove project
remove.snmfProject("genotypes_I.snmfProject")