A genotype likelihood function for DNA mixtures.

Area of Science:

• Forensic Genomics
• Population Genetics
• Statistical Genetics

Background:

• Genetic genealogy has increased demand for genome-scale forensic analyses, particularly kinship estimation.
• Current kinship estimators primarily use single nucleotide polymorphisms (SNPs) and struggle with mixed DNA samples.
• Genotype uncertainty and lack of genotype likelihood expressions for imbalanced mixtures hinder analysis of complex samples.

Purpose of the Study:

• To develop a genotype likelihood expression for imbalanced DNA mixtures.
• To integrate this expression into a genomic pipeline for kinship estimation from mixed samples.
• To assess the feasibility of deconvolving two-person mixtures using the new method.

Main Methods:

• Developed a novel genotype likelihood formulation for imbalanced mixtures.
• Integrated the formulation into a pipeline involving genotype imputation.
• Assessed mixture deconvolution through simulations and in silico experiments.

Main Results:

• The new formulation is compatible with existing genotype imputation software.
• Simulations show that read depths of 60-190 can deconvolve two-person mixtures with moderately imbalanced fractions.
• Certain mixture scenarios and SNP types remain challenging for deconvolution.

Conclusions:

• The developed approach provides a viable method for kinship estimation in mixed DNA samples.
• This work advances forensic genomics by enabling analysis of previously intractable sample types.
• Further research can refine mixture deconvolution and expand applications in forensic investigations.