HardyWeinberg (version 1.7.8)

HWChisq: Chi square tests for Hardy Weinberg equilibrium

Description

HWChisq performs the chi-square test for Hardy-Weinberg equilibrium both for autosomal and X-chromosomal markers; it can dealt both with bi-allelic and multi-allelic variants.

Usage

HWChisq(X, cc = 0.5, verbose = TRUE, x.linked = FALSE, phifixed = NULL)

Value

HWChisq returns a list with the components:

chisq

value of the chi-square statistic. NA is returned if the marker is monomorphic.

pval

p-value of the chi-square test for Hardy-Weinberg equilibrium.

D

Half the deviation from Hardy-Weinberg equilibrium for the AB genotype.

p

the allele frequency of A.

f

the inbreeding coefficient.

expected

the expected counts under Hardy-Weinberg equilibrium.

chi.contrib

the contributions of the different genotypes to the chi-square statistic.

Arguments

X

For bi-allelic variants, X is a vector containg the genotypic counts (AA,AB,BB for autosomal markers c(A,B,AA,AB,BB) for X-chromosomal markers). For multi-allelic variants, X is a lower triangular matrix with genotype counts, homozygotes on the diagonal and heterozygotes below the diagonal.

cc

cc the continuity correction parameter, the correction is only applied to bi-allelic markers (default cc = 0.5).

verbose

verbose = TRUE prints results, verbose = FALSE is silent.

x.linked

x.linked=FALSE indicates the marker is autosomal (default), and x.linked=TRUE indicates it resides on the X-chromosome.

phifixed

(For X-chromosomal markers only) phifixed=NULL indicates that the fraction of males (females) should be estimated from the data (default). If set to any other value (e.g. phifixed=0.5) then the sample is assumed to come from a population with the specified fraction of males.

Author

Jan Graffelman

Details

HWChisq does a chi-square test for Hardy-Weinberg equilibrium, and by default applies a continuity correction. For extreme allele frequencies, the continuity correction can lead to excessive type 1 error rates, and is better turned off in that case. The continuity correction can be turned off by specifying cc=0.

HWChisq can do the chi-square test for both autosomal and X-chrosomal markers. By setting x.linked = TRUE the marker will be assumed to be on the X-chromosome, and the count vector supplied should have 5 elements instead of 3 elements for an autosomal marker. For X-chromsomal markers argument phifixed is in general best left to its default value (NULL). Only in specific situations where the theoretical population sex ratio is known (e.g. in simulation studies where a universe with known gender ratio is sampled) phifixed could be set to the theoretical ratio of interest.

With bi-allelic variants, when alternative is set to less, a one-sided test for against a negative inbreeding coefficient (heterozygote excess) is performed. When alternative is set to greater a one-sided test for against a positive inbreeding coefficient (lack of heterozygotes) is performed.

For multi-allelic variants, which typically do have some rare alleles and rare genotypes, the asymptotic chi-square test is in general not recommended, and exact test procedures or permutation tests are recommended (see HWExact or HWPerm.mult).

References

Weir, B.S. (1996) Genetic data analysis II. Sinauer Associates, Massachusetts. See Chapter3.

For the chi-square test for X-linked markers:

Graffelman, J. & Weir, B.S. (2016) Testing for Hardy-Weinberg equilibrium at bi-allelic genetic markers on the X chromosome. Heredity 116(6) pp. 558--568. tools:::Rd_expr_doi("10.1038/hdy.2016.20")

See Also

HWLratio, HWChisqStats

Examples

#
# Test for an autosomal blood group marker
#
x <- c(MM=298,MN=489,NN=213)
HW.test <- HWChisq(x,verbose=TRUE)
#
# Same test without continuity correction
#
HW.test <- HWChisq(x,cc=0,verbose=TRUE)
#
# Test for X-chromsomal SNPs.
#
rs5968922 <-  c(A=392, B=212, AA=275, AB=296, BB=80)
HW.test <- HWChisq(rs5968922,cc=0,x.linked=TRUE,verbose=TRUE)
#
#
#
y <- c(GG=48, CG=209, CC=277, G=129, C=337) 
HWChisq(y,x.linked=TRUE,cc=0)
#
# Test a multi-allelic microsatellite
#
data(NistSTRs)
A1 <- NistSTRs[,1]
A2 <- NistSTRs[,2]
GC <- AllelesToTriangular(A1,A2)
HW.test <- HWChisq(GC)
#
# retaining only the three  most common alleles
#
ii <- (A1 == 10 | A1 == 11 | A1 == 12) & (A2 == 10 | A2 == 11 | A2 == 12)
A1s <- A1[ii]
A2s <- A2[ii]
GC <- AllelesToTriangular(A1s,A2s)
HW.test <- HWChisq(GC)