GWAS & Selection signatures

Genetic variation of quantitative traits modelled with full multilocus interaction

---

Abstract

Our understanding of biochemical pathways leads us to expect that interaction of genes would be the norm. The estimates of epistatic genetic variation in multifactorial traits rarely reach high values when found using family or genomic information in plant, livestock or human populations. This may be explained by low heterozygosity due to typical allele frequency distributions or by limited numbers of interacting loci, with both resulting in a major role for the additive type of genetic variation (Va). The research objective is to respond to the criticism about using a narrow range of interacting loci and to assess what kind of variation is seen when extended number of loci affect the variation. Also the properties of the respective genotypic distributions are investigated.The expected genetic variation components are found by constructing a population mean in a random mating population and deriving the main and interaction effects (or additive effects and their deviations) of the genes by the Kojima method. Va and epistatic variances (Vaa, Vaaa, etc.) are computed for 5, 10 and 50 biallelic loci with the same allele frequency 0.5 (i.e. maximum heterozygosity). The genotypic values at a locus are 2a, a and 0 (additive gene action) and the interaction effect over two, three etc. loci is [aa], [aaa], etc. Interactions are assumed to be of the biologically meaningful synergistic nature. Results are validated by independent analytical methods. The variances are expressed as proportion of the overall genetic variance. Assuming equal single locus and interaction effects, then as the number of loci increases, the proportion of Va decreases; with a large number of loci the high order interaction effects contribute the major share of the variation (Table 1a). This is accompanied by an unrealistic high skewness (over 10) of genotypic distribution. When effects are reciprocal in the locus number (n), e.g. a = n-1/2, [aa] = n-1, [aaa] = n-3/2 and so on, the proportion of Va stays very large and the high order epistatic variances are negligible (Table 1b). The distribution skewness remains in a reasonable range (0.094 to 1.50). Further, if the loci have equal effects with the highest order of interaction is restricted to five, then with more than 50 loci the proportion of Va is over 0.93 and the distribution skewness is less than 1.23.Conclusions: Despite the common presence of gene interaction, the observed high values of Va and symmetric phenotypic distributions can be explained by graduated interaction effects across large number of loci. Therefore in multilocus models a modest level of interaction would be sufficient to assess the expected types and amounts of genetic variation.

Keywords: 2026

How to Cite:

Mäki-Tanila, A. & Ganesalingam, M., (2026) “Genetic variation of quantitative traits modelled with full multilocus interaction”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2280562. doi: https://doi.org/10.31274/wcgalp.23447

Rights: 1