TECHNICAL ADVANCES: A maximum-likelihood relatedness estimator allowing for negative relatedness values

Area of Science:

• Quantitative genetics
• Statistical genetics
• Bioinformatics

Background:

• Accurate estimation of pairwise relatedness (r) is crucial in genetic studies.
• Existing maximum-likelihood estimators have limitations, particularly regarding negative relatedness estimates.
• The regression interpretation of relatedness offers an alternative framework.

Purpose of the Study:

• To extend the maximum-likelihood pairwise relatedness (r) estimator to accommodate negative estimates.
• To evaluate the performance of the new estimator against existing methods.
• To investigate the impact of allele frequency estimation errors on relatedness estimators.

Main Methods:

• Established equivalency between likelihoods in kinship programs and Thompson's likelihoods.
• Developed a new maximum-likelihood (ML) estimator for pairwise relatedness.
• Evaluated estimator performance using Monte Carlo simulations.
• Assessed the effects of allele frequency estimation errors.

Main Results:

• The new ML estimator demonstrated significantly faster unbiasedness with increased genotype information compared to the original M estimator.
• The extension successfully allowed for negative r estimates under the regression interpretation.
• Allele frequency estimation errors were considered for both new and existing estimators.

Conclusions:

• The enhanced ML estimator provides a more robust and accurate method for assessing pairwise relatedness.
• This improved estimator is particularly beneficial when dealing with substantial genotype data.
• The study highlights the importance of considering allele frequency estimation errors in relatedness analyses.