The power of the N-test of haplotype concordance

Disease susceptibility genes and the sib-pair method: a review of recent methodology

There is now a considerable body of literature which describes and analyses various methods of testing for the presence of disease susceptibility genes by examining the degree of HLA haplotype sharing of parental haplotypes over random expectations amongst affected siblings. We here survey a number of improvements, generalizations and extensions of the earlier reported methods with the aim of bringing them to the attention of those who are collecting HLA or other haplotype data on familial disease.

The distribution of Q: a powerful sibship test of association

Non-random inheritance of the two parental haplotypes among siblings affected by certain diseases has long been used to provide evidence of the presence of disease susceptibility genes. The distribution of a powerful test, called Q, based on haplotype concordance and discordance, is derived under the null hypothesis of random inheritance of haplotypes by affected siblings. The presence of a disease gene possibly linked to one of the haplotypes causes a change in the distribution of these haplotypes in the affected siblings. This distribution is found to be that of the sum of two independent variates contributed by the two parents in all parental types except one where both parents are heterozygous for the disease allele; which is dealt with separately. For comparison, the tables showing the powers of the test along with those of another well-known test, the N-test of haplotype concordance, are given. This is because Q is a modification of N to deal with the case when information on unaffected siblings is also available.

Power comparison of various sibship tests of association

There are several tests available for use with family studies to test if there is significant evidence of association with a particular disease gene locus, especially of HLA. We have used large simulations to compare the powers of tests using criteria N, R, T, PR, WP and some new tests involving both affecteds and unaffecteds in a sibship. It was found that tests using both affected and unaffected siblings were often more powerful.