A new framework to test parent-child and full sibling relationships with population substructure

Area of Science:

• Forensic Genetics
• Statistical Genetics
• Population Genetics

Background:

• Familial searching uses statistical methods to infer genetic relationships between forensic DNA samples and database individuals.
• Existing likelihood ratio-based statistics for parent-child and sibling relationships are affected by population substructure.

Purpose of the Study:

• To propose and evaluate three novel statistical methods for familial searching in DNA databases.
• To assess the performance of these statistics concerning Type I error rates and statistical power.
• To investigate the impact of population substructure on familial relationship testing.

Main Methods:

• Development of three new statistics: local allele frequency, Balding-Nichols approach, and maximum genotype probability ratio across racial subgroups.
• Performance evaluation based on Type I error and power analyses.
• Simulation studies conducted on original and extended CODIS core loci.

Main Results:

• The three proposed statistics demonstrated effectiveness in testing parent-child and full sibling relationships.
• Statistical power varied across racial groups, being highest for African-American and lowest for Caucasian samples.
• Extended CODIS core loci showed substantially higher power compared to original loci in simulation studies.

Conclusions:

• The novel statistical methods offer improved accuracy for familial searches in forensic DNA databases.
• The choice of genetic loci significantly impacts the power of familial searching, with extended CODIS loci being more effective.
• Understanding racial group differences in power is crucial for interpreting familial search results.