Determination of sample size for testing associations between genetic markers and quantitative traits in trait-based analysis

Abstract

A general formula for computing the required sample size for DNA genotyping was developed for between-population sampling schemes (control vs. selected lines in one-way selection) and within-population sampling schemes (two-tail sampling, tri-sampling or multi-sampling). In DNA fingerprinting (presence or absence of a band), the minimum sample size required for detection of marker-trait association depends upon three factors: 1) the level of significance (α = 0.05 or 0.01) and degrees of freedom for χ 2 values: a higher level of significance and a greater d.f. needs a greater sample size; 2) the sum of squares in marker frequencies between groups: a greater sum of squares requires a smaller sample size; and 3) the product of [Formula: see text], where [Formula: see text] is the average of marker genotypic frequencies among groups. The product is maximum when [Formula: see text]. A larger product requirse a greater sample size. The proposed tri-sampling allows for the detection of gene action of the linked QTL, but requires a larger sample size than two-tail sampling. Detection of non-additive gene action requires a smaller sample size than the detection of additive gene action in tri-sampling scheme. The required sample size increases rapidly with increasing number of groups sampled in trait-based analysis. The required sample size is also derived for RFLP genotyping of a diallelic locus (three marker genotypes: +/+, +/−, and −/−) and a multiallelic locus. The restriction fragment length polymorphism (RFLP) genotyping requires a smaller sample size than DNA fingerprinting for detection of marker-QTL association. Key words: Sample size, genetic markers, trait-based analysis