genomicrelatedness

Genomicrelatedness is the degree of genetic similarity between individuals as estimated from genome-wide genetic data. It captures realized relatedness arising from shared ancestry and Mendelian sampling, and it complements pedigree-based kinship estimates by using actual observed genetic variants rather than solely expecting relationships from pedigrees. Genomicrelatedness can be positive for individuals who share recent ancestry and near zero or negative for those with little recent shared ancestry, depending on the calculation method and reference allele frequencies.

A common way to quantify genomicrelatedness is through a genomic relationship matrix (GRM) or realized relationship matrix. These matrices are typically constructed from single-nucleotide polymorphism (SNP) data by centering and scaling genotypes according to allele frequencies, producing pairwise relatedness estimates. Methods such as the VanRaden approach and related implementations are widely used, sometimes distinguishing identity by descent from identity by state. Software tools like GCTA, PLINK, and KING implement practical workflows for estimating GRMs and related statistics from large datasets.

Applications of genomicrelatedness are broad. In breeding programs, it underpins genomic selection and estimation of genetic merit. In human and population genetics, it helps control for population structure and cryptic relatedness in genome-wide association studies, facilitates SNP-based heritability estimation, and informs inbreeding and demographic history analyses. It also supports quality control and kinship inference in sequencing and genotyping datasets.

Limitations include sensitivity to allele frequency estimates, population structure, and admixture, which can bias relatedness inferences if reference frequencies are inappropriate. Accurate estimation requires careful quality control, adequate sample size, and appropriate handling of missing data and imputation.