Searching national DNA databases with complex DNA profiles: An empirical study using probabilistic genotyping.

Area of Science:

• Forensic Science
• Genetics
• Computational Biology

Background:

• National DNA databases typically only search single-source or two-person DNA mixtures.
• Complex DNA profiles (e.g., 3+ contributors) often cannot be searched, limiting their investigative utility.
• Probabilistic genotyping (PG) offers a method to analyze and compare complex DNA mixtures.

Purpose of the Study:

• To evaluate the performance of the DBLR™ "Search database" tool with the Swiss National DNA Database (NDNADB).
• To assess the effectiveness of PG in searching complex DNA mixtures against a large DNA database.
• To determine the sensitivity and specificity of the DBLR™ tool for identifying true and adventitious associations.

Main Methods:

• Preparation and deconvolution of 40 artificial DNA mixtures (2-5 contributors) using STRmix™.
• Comparison of deconvoluted mixtures against a database of 15 volunteers and 174,493 Swiss NDNADB profiles (ground-truth).
• Analysis of 160 casework DNA mixture profiles (2-4 contributors) using DBLR™ and likelihood ratio (LR) thresholds (10^3 and 10^6).

Main Results:

• Ground-truth experiments achieved high sensitivity (90.0%–99.9%) and specificity (57.1%–100.0%) using LR thresholds of 10^3 and 10^6.
• Casework profile searches with a 10^3 LR threshold retrieved 380 associations, including 194 expected and 186 new leads.
• Even complex mixtures (up to 4 contributors) yielded all expected associations, generating new investigative leads.

Conclusions:

• Database searches using probabilistic genotyping for complex DNA mixtures can generate crucial investigative leads.
• The DBLR™ tool shows promise for enhancing the utility of complex DNA profiles in forensic investigations.
• Further discussions on legal, financial, and technical aspects are needed for operational integration of such tools.