Decision analysis for the genotype designation in low-template-DNA profiles.

Area of Science:

• Forensic Science
• Genetics
• Decision Theory

Background:

• Forensic scientists often analyze low-template-DNA (lt-DNA) traces, presenting challenges for accurate genotype determination.
• Database searches for identifying the source of lt-DNA require careful genotype specification to minimize errors.
• Current methods may lack a standardized, decision-driven approach for genotype selection in lt-DNA analysis.

Purpose of the Study:

• To develop a normative, decision-theoretic framework for selecting the optimal genotype designation for lt-DNA database searches.
• To minimize potential errors, specifically false exclusions and false inclusions, in forensic DNA profiling.
• To provide a data-driven method for scientists to make informed genotype decisions on a per-locus basis.

Main Methods:

• Defined a decision-theoretic framework incorporating genotype uncertainty and a loss function for potential errors.
• Applied the framework to two hypothetical cases involving single or multiple allele peak observations across replicates.
• Derived threshold values for genotype designation switches based on observed peak heights and allele dropout probabilities.

Main Results:

• The proposed framework guides genotype selection by minimizing expected loss, thereby reducing forensic database search errors.
• Threshold values for switching genotype designations were derived and shown to be dependent on observed peak heights.
• Sensitivity analyses demonstrated the influence of model parameters on these critical threshold values.

Conclusions:

• A decision-theoretic approach offers a robust method for optimizing genotype specification in lt-DNA database searches.
• The optimal genotype designation strategy is locus- and laboratory-dependent, necessitating flexible threshold values.
• This framework enhances the reliability and accuracy of forensic DNA analysis, particularly for challenging trace samples.