Efficient construction of match strength distributions for uncertain multi-locus genotypes

Area of Science:

• Forensic Science
• Computational Biology
• Biostatistics

Background:

• Natural variation in biological evidence creates uncertain DNA genotypes.
• Forensic DNA comparisons yield numerical strength of association, but multi-locus analysis is computationally challenging.

Purpose of the Study:

• To develop an efficient method for constructing multi-locus DNA match strength distributions.
• To accurately quantify the information content of complex DNA evidence.

Main Methods:

• Utilizing Bayesian probability to generate locus-specific match strength random variables.
• Employing convolution to combine independent locus distributions into a multi-locus distribution.
• Demonstrating function convergence for accuracy and quadratic computational complexity.

Main Results:

• Efficiently constructed multi-locus match strength distributions in fractions of a second.
• Achieved high accuracy with increased numerical resolution.
• Reduced high-dimensional computational cost to fast real-line arithmetic.

Conclusions:

• Convolution construction rapidly and accurately generates multi-locus DNA match strength distributions.
• This method enhances the forensic validation of DNA evidence and error rate assessment.
• The approach is applicable to various scientific fields involving discrete or continuous variables.