Bleka, O. | bioRxiv Science

bioRxiv2026

Activity-level evaluation of DNA transfer to stockings: comparisonof HaloGen with simplified contributor-based approaches

DNA findings on clothing are often evaluated at sub-source level to address whether a person contributed DNA to a sample. In many cases, however, the disputed issue is not whose DNA is present, but how the DNA was transferred. This paper applies the HaloGen framework to evaluate DNA findings on stockings under activity-level propositions. The data derive from controlled experiments, reported in a companion paper, comparing active contact, in which a person of interest pulled down a pair of stockings, with social contact, in which the person of interest and wearer used the same bathroom. Contributor-specific DNA quantities were estimated for the person of interest and for observed unknown contributors across two sampled stains. Compatibility diagnostics indicated that the stockings dataset was broadly compatible with the observed Pop20 reference range. Pop20 was therefore used as regularising prior information in the Lab-Bayes model, rather than as a stand-alone Group model. Because the stockings dataset was generated in a single laboratory, transfer data from 20 laboratories were used as regularising prior information for a laboratory-specific Lab-Bayes model. The local stockings data then updated the direct- and secondary-transfer parameter distributions. Compatibility diagnostics indicated that the stockings dataset was broadly compatible with the observed Pop20 reference range. Pop20 was therefore used as regularising prior information in the Lab-Bayes model, rather than as a stand-alone Group model. In addition, prior-sensitivity and leave-one-case-out analyses indicated that the case-level likelihood ratios were not materially affected by the choice of prior specification or by data reuse. HaloGen likelihood ratios were assessed using Tippett plots and compared with simplified contributor-based comparator approaches. These approaches represent simplified forms of activity-level reasoning in which the DNA result is reduced to binary or categorical states. They comprise a binary source-support approach, a discrete Mx component-size approach, and a background approach based on the ReAct I Experiment 3 formulation. These approaches were used to examine the effect of reducing continuous contributor-specific DNA quantities to binary or categorical states, and to assess how the likelihood ratio is affected when observed unknown contributors are represented through a background/direct-unknown term rather than evaluated explicitly as possible alternative direct actors. HaloGen showed clear discrimination between active-contact and social-contact ground-truth cases. The simplified comparator methods illustrated the information loss caused by thresholding or categorising continuous DNA quantities. The ReAct background approach tended to shift direct-transfer likelihood ratios upward relative to the HaloGen calculation, showing that the treatment of observed unknown contributors can materially affect activity-level inference. The results support the importance of preserving contributor-level symmetry: an observed unknown contributor should not automatically be demoted to background when that contributor may represent an alternative direct actor under the defence proposition.

genetics↗

bioRxiv2026

Likelihood Ratios Given Activity-Level Propositions for DNA Transfer Evidence: Practical Implementation and Simulation Studies Using the HaloGen Engine (Part II)

The quantitative interpretation of low-template DNA findings given activity-level propositions requires models that can accommodate inter-laboratory variability, uncertainty in transfer and recovery, and case-specific assumptions. This paper presents the practical implementation of HaloGen, an open-source hierarchical Bayesian framework for calculating activity-level likelihood ratios (LRs) from DNA quantity data. We compare three modelling approaches derived from the framework: a Group model, which combines data across laboratories, a hierarchically informed Lab-Bayes model, and a standalone, laboratory specific Lab-Vague model. Through simulation studies, we show that evidential strength is sensitive not only to DNA quantity but also to case context, particularly the assumed number of relevant actors (NS), the treatment of specified unknown contributors, and the choice of laboratory calibration. Inter-laboratory differences in DNA recovery and non-detection can lead to materially different LRs when these data are used within the HaloGen framework, so pooled or external data should not be used uncritically. To address practical implementation, we propose a minimum-effort calibration pathway for laboratories wishing to use HaloGen for quantitative activity-level LR reporting. The results indicate that a limited number of local direct/secondary transfer experiments can improve relevance compared with exclusive reliance on a pooled population model, although the adequacy of any dataset remains case- and proposition-dependent. The findings clarify how contextual assumptions enter mathematically into activity-level inference and underscore the importance of transparent specification of propositions, data relevance, model assumptions, and remaining expert judgement.

genetics↗

bioRxiv2026

Likelihood Ratios Given Activity-Level Propositions for DNA Transfer Evidence: Theoretical Foundations of the HaloGen Framework (Part I)

The interpretation of trace DNA evidence at activity level requires explicit modelling of transfer, persistence, and the possibility that a relevant actor leaves no detectable DNA. We present the theoretical foundations of Halo-Gen, an open-source hierarchical Bayesian framework for evaluating DNA quantities under competing activity-level propositions. The propositions are framed in terms of alleged activities or actions; direct and secondary transfer are treated as transfer mechanisms through which those activities may give rise to the observed DNA. HaloGen accounts for zero transfer, multiple contributors, specified unknown contributors, unobserved actors, and multiple stains. Evidence is evaluated using an exhaustive-propositions likelihood-ratio framework that combines information across contributors and stains while propagating uncertainty in transfer and detection. Observed DNA quantities and non-detects are handled within a single probabilistic framework: detected quantities are evaluated using conditional-on-detection densities, whereas the probability that a relevant actor leaves no detectable DNA is represented by an empirically constrained fail-rate parameter. The framework yields transparent and stable behaviour: informative DNA quantities can support propositions involving direct transfer, while low-information or no-detect situations are neutral or defence-conservative. The empirical-clamped fail-rate policy prevents spurious inflation of likelihood ratios when non-detection of a relevant actor is plausible. This paper establishes the theoretical basis of HaloGen; a companion paper addresses validation and applied casework examples.

genetics↗

bioRxiv2024

Low LRs obtained from DNA mixtures: On calibration and discrimination performance of probabilistic genotyping software

The validity of a probabilistic genotyping (PG) system is typically demonstrated by following international guidelines for the developmental and internal validation of PG software. These guidelines mainly focus on discriminatory power. Very few studies have reported with metrics that depend on calibration of likelihood ratio (LR) systems. In this study, discriminatory power as well as various calibration metrics, such as Empirical Cross-Entropy (ECE) plots, pool adjacent violator (PAV) plots, log likelihood ratio cost (Cllr and Cllrcal), fiducial calibration discrepancy plots, and Turing expectation were examined using the publicly-available PROVEDIt dataset. The aim was to gain deeper insight into the performance of a variety of PG software in the lower LR ranges ([~]LR 1-10,000), with focus on DNAStatistX and EuroForMix which use maximum likelihood estimation (MLE). This may be a driving force for the end users to reconsider current LR thresholds for reporting. In previous studies, overstated low LRs were observed for these PG software. However, applying (arbitrarily) high LR thresholds for reporting wastes relevant evidential value. This study demonstrates, based on calibration performance, that previously reported LR thresholds can be lowered or even discarded. Considering LRs >1, there was no evidence for miscalibration performance above LR [~]1,000 when using Fst 0.01. Below this LR value, miscalibration was observed. Calibration performance generally improved with the use of Fst 0.03, but the extent of this was dependent on the dataset: results ranged from miscalibration up to LR [~]100 to no evidence of miscalibration alike PG software using different methods to model peak height, HMC and STRmix. This study demonstrates that practitioners using MLE-based models should be careful when low LR ranges are reported, though applying arbitrarily high LR thresholds is discouraged. This study also highlights various calibration metrics that are useful in understanding the performance of a PG system. HighlightsO_LIDiscriminatory power and calibration performance of PG software are evaluated. C_LIO_LIThe utility of various calibration metrics are explored in low LR ranges. C_LIO_LIFocus was on DNAStatistX and EuroForMix software using the MLE method. C_LIO_LICalibration performance was dependent on Fst value and dataset size. C_LIO_LIResults suggest reconsideration of lower LR thresholds and cautious reporting of low LRs. C_LI

genetics↗

bioRxiv2022

Limitations of qPCR to estimate DNA quantity: AnRFU method to facilitate inter-laboratory comparisonsfor activity level, and general applicability

The application of qPCR to estimate the quantity of DNA present is usually based upon a short amplicon (typically c.80bp) and a longer amplicon (typically c.200-300bp) where the latter is used to determine the amount of degradation present in a sample. The data are used to make decisions about a) whether there is sufficient template to amplify? b) how much of the elution volume to forward to PCR? A typical multiplex amplifies template in the region of 100-500bp. Consequently, the results from an 80bp amplicon will tend to overestimate the actual amplifiable quantity that is present in a degraded sample. To compensate, a method is presented that relates the quantity of amplifiable DNA to the average RFU of the amplified fragments. This provides greatly improved accuracy of the estimated quantity of DNA present, which may differ by more than an order of magnitude compared to qPCR. The relative DNA quantities can be apportioned per contributor once mixture proportions are ascertained with probabilistic genotyping software (EuroForMix). The motivation for this work was to provide an improved method to generate data to prepare distributions that are used to inform activity level propositions. However, other applications will benefit, particularly those where extraction and quantification are bypassed: For example direct PCR and Rapid DNA technology. The overall aim of this work was to provide a method of quantification that is standardised and can be used to compare results between different laboratories that use different multiplexes. A software solution "ShinyRFU" is provided to aid calculations.

biochemistry↗

bioRxiv2021

RFU derived LRs for activity level assignments usingBayesian Networks

The application of qPCR to estimate the quantity of DNA present is usually based upon a short amplicon (typically c.80bp) and a longer amplicon (typically c.200-300bp) where the latter is used to determine the amount of degradation present in a sample. The data are used to make decisions about a) whether there is sufficient template to amplify? b) how much of the elution volume to forward to PCR? A typical multiplex amplifies template in the region of 100-500bp. Consequently, the results from an 80bp amplicon will tend to overestimate the actual amplifiable quantity that is present in a degraded sample. To compensate, a method is presented that relates the quantity of amplifiable DNA to the average RFU of the amplified fragments. This provides greatly improved accuracy of the estimated quantity of DNA present, which may differ by more than an order of magnitude compared to qPCR. The relative DNA quantities can be apportioned per contributor once mixture proportions are ascertained with probabilistic genotyping software (EuroForMix). The motivation for this work was to provide an improved method to generate data to prepare distributions that are used to inform activity level propositions. However, other applications will benefit, particularly those where extraction and quantification are bypassed: For example direct PCR and Rapid DNA technology. The overall aim of this work was to provide a method of quantification that is standardised and can be used to compare results between different laboratories that use different multiplexes. A software solution "ShinyRFU" is provided to aid calculations.

genetics↗

bioRxiv2021

An LR framework incorporating sensitivity analysis to model multiple direct and secondary transfer events on skin surface

Bayesian logistic regression is used to model the probability of DNA recovery following direct and secondary transfer and persistence over a 24 hour period between deposition and sample collection. Sub-source level likelihood ratios provided the raw data for activity-level analysis. Probabilities of secondary transfer are typically low, and there are challenges with small data-sets with low numbers of positive observations. However, the persistence of DNA over time can be modelled by a single logistic regression for both direct and secondary transfer, except that the time since deposition must be compensated by an offset value for the latter. This simplifies the analysis. Probabilities are used to inform an activity-level Bayesian Network that takes account of alternative propositions e.g. time of assault and time of social activities. The model is extended in order to take account of multiple contacts between person of interest and victim. Variables taken into account include probabilities of direct and secondary transfer, along with background DNA from unknown individuals. The logistic regression analysis is Bayesian -for each analysis, 4000 separate simulations were carried out. Quantile assignments enable calculation of a plausible range of probabilities and sensitivity analysis is used to describe the corresponding variation of LRs that occur when modelled by the Bayesian network. It is noted that there is need for consistent experimental design, and analysis, to facilitate inter-laboratory comparisons. Appropriate recommendations are made. The open-source program written in R-code ALTRaP (Activity Level, Transfer, Recovery and Persistence) enables analysis of complex multiple transfer propositions that are commonplace in cases-work e.g. between those who cohabit. A number of case examples are provided. ALTRaP can be used to replicate the results and can easily be modified to incorporate different sets of data and variables.

genetics↗