EuroForMix

EuroForMix implements a continuous-model statistical framework for quantitative interpretation and likelihood-ratio evaluation of STR (Short Tandem Repeat) DNA mixture profiles in forensic genetics.

Key Features:

• Continuous peak-height model: Quantifies allelic peak heights on a continuous scale for probabilistic interpretation of electropherogram signals.
• Likelihood-ratio calculations: Computes weight-of-evidence using likelihood ratios for competing contributor hypotheses.
• Parameter inference: Performs inference on mixture proportions, peak-height properties, stutter proportion, and degradation effects.
• Artefact modeling: Models allele drop-out, allele drop-in, stutter, and population substructure within the statistical framework.
• Inference methodologies: Supports both maximum likelihood estimation (MLE) and Bayesian inference with user-specified prior distributions.
• Contributor handling: Handles models with known and unknown contributors and supports up to four unknown contributors in practice due to computing time constraints.
• Validation: Implementation has been validated against the R-package DNAmixtures (which requires a HUGIN Expert license).

Scientific Applications:

• Forensic mixture interpretation: Quantitative evaluation of mixed STR DNA profiles to support casework and evidentiary assessment.
• Weight-of-evidence reporting: Calculation of likelihood ratios for competing hypotheses about contributor identities and numbers.
• Contributor and parameter estimation: Estimation of contributor proportions, genotype inference, and estimation of stutter and degradation parameters.
• Artefact assessment and population effects: Analysis that accounts for drop-in, drop-out, stutter, and sub-population structure in STR data.
• Method comparison and benchmarking: Platform for comparing analytical solutions and validating continuous-model implementations against other methods such as DNAmixtures.

Methodology:

Uses a continuous peak-height model to quantify allelic peak heights and performs parameter inference via maximum likelihood estimation and Bayesian inference with priors, explicitly modeling stutter, allele drop-in, allele drop-out, degradation, and population substructure, with validation against the R-package DNAmixtures.