A likelihood ratio approach for identifying three-quarter siblings in genetic databases

Area of Science:

• Genetics
• Bioinformatics
• Population Genetics

Background:

• Detecting family relationships in genetic databases is crucial for genetic epidemiology, forensics, and genealogy.
• Current methods primarily identify first and second-degree relatives, overlooking intermediate relationships like three-quarter siblings.
• Increasing genetic database sizes make the presence of such intermediate relationships more probable.

Purpose of the Study:

• To extend likelihood ratio (LR) methodology for accurate inference of three-quarter sibling (3/4S) relationships.
• To differentiate 3/4S from first-degree (full siblings) and second-degree relatives within genetic datasets.
• To provide a robust method for quality control in large-scale genetic studies.

Main Methods:

• Utilized an extended likelihood ratio (LR) framework to quantify relatedness.
• Employed bootstrap confidence intervals to assess the uncertainty of LR estimations.
• Incorporated marker pruning to account for linkage disequilibrium (LD) and validated with simulations including recombination.

Main Results:

• Successfully developed and validated a method to accurately detect three-quarter siblings (3/4S).
• Demonstrated the ability to distinguish 3/4S from full siblings and second-degree relatives.
• The methodology was illustrated using empirical genome-wide data from the GCAT Genomes for Life cohort.

Conclusions:

• The proposed LR-based method effectively identifies three-quarter siblings in genetic databases.
• This advancement enhances the accuracy of family relationship detection, particularly in large genetic datasets.
• The method contributes to improved quality control and deeper insights in population and genealogical genetics.