Computationally efficient sibship and parentage assignment from multilocus marker data.

Area of Science:

• Genetics
• Computational Biology
• Bioinformatics

Background:

• Inferring familial relationships (sibship, parentage) from genetic data is crucial in various biological studies.
• Existing methods, including exclusion and likelihood-based approaches, face computational limitations with large datasets.
• Full-likelihood methods provide high accuracy but are computationally intensive, limiting their application.

Purpose of the Study:

• To develop a computationally efficient likelihood-based method for inferring sibship and parentage.
• To enable accurate relationship inference in large populations with multilocus marker genotypes.
• To overcome the computational limitations of existing full-likelihood methods.

Main Methods:

• A novel likelihood-based method utilizing the sum of log likelihoods of pairwise relationships.
• Pre-calculation and storage of pairwise relationship log likelihoods for efficiency.
• Evaluation using empirical and simulated datasets, comparing against exclusion and full-likelihood methods.

Main Results:

• The proposed method demonstrates higher accuracy than pairwise likelihood and exclusion-based methods.
• It achieves accuracy comparable to, though slightly less than, the full-likelihood method.
• The new method is significantly more computationally efficient than the full-likelihood method, especially for complex scenarios like polygamy and genotyping errors.

Conclusions:

• The new method provides a computationally feasible and accurate solution for inferring genetic relationships in large datasets.
• It offers a practical alternative to computationally prohibitive full-likelihood methods.
• The approach scales well for thousands of individuals, limited primarily by computer memory.