Salience, Legitimacy, and Credibility in Single-Cell Forensic Systems
Organized by: International Forensic Scientist Awards
Website: forensicscientist.org
14th Edition of Forensic Scientist Awards 26-27 September 2025 | Mumbai, India
In forensic investigations where no suspect is identified, national DNA databases offer a critical tool for generating leads. However, these databases have strict limitations on the type of DNA data that can be uploaded. For example, data derived from complex mixtures—more than two contributors—often cannot be used, leaving many cases unresolved.
A single-cell approach offers a promising solution. By isolating individual cells at the very start of the workflow, forensic scientists can extract DNA from each cell independently. Once DNA signatures are obtained, they are clustered into groups, and probabilities are calculated to determine which genotype most likely contributed to each cluster. Using Bayes’ Rule, this method identifies the most probable genotype for database queries. Effective clustering is thus essential for reliable single-cell forensic analysis.
In our research, we evaluated two clustering strategies within the end-to-end single-cell prediction system EESCIt™: Model-Based Clustering (MBC) and Forensic-Aware Clustering (FAC). To measure performance, we structured our analysis into three key categories: Salience, Legitimacy, and Credibility (SLC).
-
Salience reflects how relevant the technology is to forensic needs. Single-cell reports have strong applicability for generating actionable leads.
-
Legitimacy assesses how accurately the system identifies clusters. FAC consistently returned correct cluster numbers for all admixtures, outperforming MBC. With FAC, 90% of loci yielded a single credible genotype—the correct one—compared to MBC’s 84%.
-
Credibility examines how reliable the probabilistic predictions are. Using Brier Scores, FAC demonstrated better calibration and refinement, reflecting higher trustworthiness in its predictions.
By adopting FAC into EESCIt™, we created the first end-to-end single-cell probabilistic system capable of answering crucial forensic questions: how many donors are present, and who they are. This advancement bridges the “mixture gap” in DNA forensics, enhancing the ability to solve previously unresolved cases and providing forensic actors with more reliable investigative tools.
🔗 Learn more and apply at:
https://forensicscientist.org/
Nominations Open Now: Click here
–––––––––––––––––––––––––––––––––––––
Get Connected Here:
Comments
Post a Comment