Analytical thresholds are a fundamental component of capillary electrophoresis (CE)–based human short tandem repeat (STR) analysis, providing the basis for distinguishing true allelic signal from baseline instrument noise. The selection of an appropriate threshold in directly impacts the interpretation and informational value of a DNA profile, particularly in low-template DNA samples and complex mixtures where potential allele drop-out can influence interpretation, particularly when assessing the number of contributors to the sample. Despite the high degree of standardization in STR chemistries and instrumentation, considerable inter-laboratory variability in analytical threshold determination persists. This variability is not reflective of the variability in assay biochemistry, rather, they reflect a broad perspective on different methods (mathematical models) and the data employed in these determinations.  As a result, the application of different analytical thresholds can result in inconsistent results, and in some cases, discordant casework outcomes.

Here, we present THOR (ThresHOld Response program), a statistically robust and openly available framework developed to promote consistency and scientific rigor in the establishment of analytical thresholds for CE-based STR systems. THOR integrates established mathematical principles from the with empirical validation strategies as a software application for determining thresholds and evaluating datasets during laboratory validation.  With an emphasis rigor and experimental design, the program promotes data set construction, and the application of well-supported statistical methodologies using standardized terminology. Central to the framework is the recognition that baseline noise and peak height data rely on varying distributions (skew) and depart from normality. Analytical approaches that rely on untested distributional assumptions or limited descriptive statistics may provide limited results.  THOR provides a structured approach for evaluating distributional characteristics, selecting appropriate statistical models, and ensuring that analytical thresholds are fit for purpose in the context of assay sensitivity and detection enhancements. The program further underscores the importance of validation design, adequate sample size, and methodological rigor in generating defensible and reproducible threshold determinations. By promoting transparent statistical practices and harmonized terminology, THOR seeks to reduce inter-laboratory variability and strengthen the scientific foundation underlying analytical threshold implementation. This framework offers laboratories a practical and defensible approach to threshold establishment prior to deployment in routine forensic casework.

Authors

• Christian Westring (Niagara County Sheriff's Office Forensic Laboratory, United States of America)
• Brady Harrington (Alfred State, United States of America)
• Eric Best (University at Albany, United States of America)
• Kimberly Cornell (University at Albany, United States of America)