Considerations on the application of a mutation model for Y-STR interpretation.

If Y-STR profiling is to be more effective in criminal casework, the methods used to evaluate evidential weight require improvement. Many forensic scientists assign an evidential weight by estimating the number of times a Y-STR profile obtained from a questioned sample has been observed in YHRD datasets. More sophisticated models have been suggested but not yet implemented into routine casework, e.g. Andersen & Balding [1]. Mutation is inherent to STR meiosis (or inheritance) and is encountered in practice. We evaluated a mutation model that can be incorporated into a method for assigning evidential weight to Y-STR profiles, an essential part of bringing any method into practice. Since an important part of implementation to casework is communication, the article is written in an accessible format for practitioners as well as statisticians. The mutation component within the MUTEA model by Willems et al. [2] incorporates the potential for multistep mutations and a tendency for alleles to revert towards a central length, reflecting observed mutation data, e.g. [3]. We have estimated the parameters in this model and in a simplified symmetric version of this model, using sequence data from father/son pairs [4] and deep-rooted pedigrees [5]. Both datasets contain multistep mutations, which may have an effect on models based on simulations [1]. We introduce Beta-Binomial and Beta-Geometric conjugate analyses for estimating rate and step parameters for the mutation models presented here, which require only summations and multiplications. We proved mathematically that the parameters can be estimated independently. We show the importance of reporting the variability of the parameters and not only a point estimate. The parameters can be easily incorporated into statistical models, and updated sequentially as more data becomes available. We recommend fuller publication of data to enable the development and evaluation of a wider range of mutation models.

Identification of bullets fired from air guns using machine and deep learning methods.

Ballistics (the linkage of bullets and cartridge cases to weapons) is a common type of evidence encountered in criminal cases around the world. The interest lies in determining whether two bullets were fired using the same firearm. This paper proposes an automated method to classify bullets from surface topography and Land Engraved Area (LEA) images of the fired pellets using machine and deep learning methods. The curvature of the surface topography was removed using loess fit and features were extracted using Empirical Mode Decomposition (EMD) followed by various entropy measures. The informative features were identified using minimum Redundancy Maximum Relevance (mRMR), finally the classification was performed using Support Vector Machines (SVM), Decision Tree (DT) and Random Forest (RF) classifiers. The results revealed a good predictive performance. In addition, the deep learning model DenseNet121 was used to classify the LEA images. DenseNet121 provided a higher predictive performance than SVM, DT and RF classifiers. Moreover, the Grad-CAM technique was used to visualise the discriminative regions in the LEA images. These results suggest that the proposed deep learning method can be used to expedite the linkage of projectiles to firearms and assist in ballistic examinations. In this work, the bullets that were compared were air pellets fired from both air rifles and a high velocity air pistol. Air guns were used to collect the data because they were more accessible than other firearms and could be used as a proxy, delivering comparable LEAs. The methods developed here can be used as a proof-of-concept and are easily expandable to bullet and cartridge case identification from any weapon.

A ground truth data set of gas chromatography mass spectrometry (GCMS) analysed synthesised methylenedioxymethylamphetamine (MDMA).

Controlled drug samples are normally chemically analysed to determine their identity and in some cases, their purity. There are also circumstances where a more broad chemical characterisation of drug samples may also be required. This involves investigating the chemical impurities that may be present in a drug sample as a consequence of their synthesis. This impurity or drug profiling can be derived from drugs which are synthesised chemically or extracted from plant materials and then modified chemically. Impurity profiling can provide some insight into the synthetic methods used and sometimes the starting chemicals used. We report on the data generated from repetitive ( n = 18 ) synthesis of ecstasy (methylenedioxymethylamphetamine or MDMA) made by three different synthetic methods. Each data sample is expressed in multiple formats. This article uses the template for publishing GCMS data provided in Miller et al.(2022)[1]. The template provides a robust and systematic approach to organise GCMS data that is both useful for practitioners and amenable for automated data manipulation by data scientists.

A UK-based ground truth data set of GCMS analysed ignitable liquid samples - a template for making chromatographic data accessible as an open source data set.

Fire debris is often recovered as part of a fire scene investigation to determine whether an ignitable liquid might be present which may be evidence of a deliberate fire. The analysis of fire debris produces chromatograms that a forensic chemist uses to determine whether or not an ignitable liquid may be present. Currently there are very few publicly available data sets that can be used for training and statistical modelling in this area. The data set in this paper has been prepared with these two applications in mind and covers a wide range of ignitable liquids available in the UK. We created a data set of 35 ignitable liquids including petrol (gasoline), light, medium and heavy petroleum distillates (i.e diesel) from several retailers. Each ignitable liquid was systematically evaporated to produce six additional samples. Each sample was repetitively analysed to provide an overall data set of 751 analytical outputs (including chromatograms). Each data sample is expressed in multiple formats and the metadata containing any data used in the production of the samples is included. The folder and file names are designed to avoid misplacements and to manipulate folders and files systematically using computer code.

Calculation of likelihood ratios for inference of biological sex from human skeletal remains.

It is common in forensic anthropology to draw inferences (e.g., inferences with respect to biological sex of human remains) using statistical models applied to anthropometric data. Commonly used models can output posterior probabilities, but a threshold is usually applied in order to obtain a classification. In the forensic-anthropology literature, there is some unease with this "fall-off-the-cliff" approach. Proposals have been made to exclude results that fall within a "zone of uncertainty", e.g., if the posterior probability for "male" is greater than 0.95 then the remains are classified as male, and if the posterior probability for "male" is less than 0.05 then the remains are classified as female, but if the posterior probability for "male" is between 0.05 and 0.95 the remains are not classified as either male or female. In the present paper, we propose what we believe is a simpler solution that is in line with interpretation of evidence in other branches of forensic science: implementation of the likelihood-ratio framework using relevant data, quantitative measurements, and statistical models. Statistical models that can implement this approach are already widely used in forensic anthropology. All that is required are minor modifications in the way those models are used and a change in the way practitioners and researchers think about the meaning of the output of those models. We explain how to calculate likelihood ratios using osteometric data and linear discriminant analysis, quadratic discriminant analysis, and logistic regression models. We also explain how to empirically validate likelihood-ratio models.

Interpretation of DNA data within the context of UK forensic science - investigation.

This article is the second part of a review of the interpretation of DNA data in forensic science. The first part describes the evaluation of autosomal profile for criminal trials where an evidential weight is assigned to the profile of a person of interest (POI) and a crime-scene profile. This part describes the state of the art and future advances in the interpretation of forensic DNA data for providing intelligence information during an investigation. Forensic DNA is crucial in the investigative phase of an undetected crime where a POI needs to be identified. A sample taken from a crime scene is profiled using a range of forensic DNA tests. This review covers investigation using autosomal profiles including searching national and international crime and reference DNA databases. Other investigative methodologies described are kinship analysis; familial searching; Y chromosome (Y-STR) and mitochondrial (mtDNA) profiles; appearance prediction and geographic ancestry; forensic genetic genealogy; and body identification. For completeness, the evaluation of Y-STRs, mtDNA and kinship analysis are briefly described. Taken together, parts I and II, cover the range of interpretation of DNA data in a forensic context.

Interpretation of DNA data within the context of UK forensic science - evaluation.

Forensic DNA provides a striking contribution to the provision of justice worldwide. It has proven to be crucial in the investigative phase of an unsolved crime where a suspect needs to be identified, e.g. from a DNA database search both nationally and internationally. It is also a powerful tool in the assignment of evidential weight to the comparison of a profile of a person of interest and a crime scene profile. The focus of this document is the evaluation of autosomal profiles for criminal trials in the UK. A separate review covers investigation and evaluation of Y-STR profiles, investigation using autosomal profiles, kinship analysis, body identification and Forensic Genetic Genealogy investigations. In less than 40 years, forensic DNA profiling has developed from a specialist technique to everyday use. Borrowing on advances in genome typing technology, forensic DNA profiling has experienced a substantial increase in its sensitivity and informativeness. Alongside this development, novel interpretation methodologies have also been introduced. This document describes the state of the art and future advances in the interpretation of forensic DNA data.

Adventitious match probability for autosomal profiles when primer binding site mutation is possible.

This paper considers the situation where two DNA systems with differing primers have been used to produce DNA profiles for loading and searching of a DNA Database. With any profiling system there exists the possibility of a "primer binding site mutation" (PBSM). When such a mutation occurs at one of the loci in a profile, it has the effect that the associated allele is not visible in the profile. In the case where a person has two different alleles at a given locus (heterozygous) the effect of a PBSM would be that the profile would appear to be that of an individual with only one allele at that locus (homozygous). The paper investigates the potential for an adventitious match as a result of a PBSM when, for example, a crime profile and person profile that have originated from two different individuals are found to be the same as a result of a PBSM in one of the profiles. It is demonstrated, both by theory and using simulations, that the effect of PBSMs is to slightly decrease the adventitious match probability from what it would had the same DNA system been used.

Providing scientific guidance on DNA to the judiciary.

A series of short documents have been written in response to a request from the UK Judiciary for explanations of research that was commissioned in response to questions they had raised. These related principally to the potential impact of primer binding site mutation (PBSM) but it became clear at an early stage that it was necessary to explain related issues. The three scientific guidance papers (SGPs) that have been prepared thus far are presented in their entirety so that UK scientists may be aware of what has been presented to judges. Suggestions for further work, including possible communication to jurors are discussed.

Evidential evaluation of DNA profiles using a discrete statistical model implemented in the DNA LiRa software.

The high sensitivity of the technology for producing profiles means that it has become routine to produce profiles from relatively small quantities of DNA. The profiles obtained from low template DNA (LTDNA) are affected by several phenomena which must be taken into consideration when interpreting and evaluating this evidence. Furthermore, many of the same phenomena affect profiles from higher amounts of DNA (e.g. where complex mixtures has been revealed). In this article we present a statistical model, which forms the basis of software DNA LiRa, and that is able to calculate likelihood ratios where one to four donors are postulated and for any number of replicates. The model can take into account dropin and allelic dropout for different contributors, template degradation and uncertain allele designations. In this statistical model unknown parameters are treated following the Empirical Bayesian paradigm. The performance of LiRa is tested using examples and the outputs are compared with those generated using two other statistical software packages likeLTD and LRmix. The concept of ban efficiency is introduced as a measure for assessing model sensitivity.

A dropin peak height model.

The technology for producing DNA profiles is very sensitive and is able to produce profiles from process negative controls typically consisting of one or two peaks, a phenomenon known as dropin. There are several types of models that implement likelihood ratios for the statistical evaluation of DNA profiles. One of the types is the family of continuous models because they consider continuous peak height/area measurements from the stain profile. Nowadays, there are several continuous models available. The aim of this article is to expose a statistical model for dropin peak heights supported by data, and illustrate its incorporation into a continuous method.

Evaluating forensic DNA profiles using peak heights, allowing for multiple donors, allelic dropout and stutters.

Increases in the sensitivity of DNA profiling technology now allow profiles to be obtained from smaller and more degraded DNA samples than was previously possible. The resulting profiles can be highly informative, but the subjective elements in the interpretation make it problematic to achieve the valid and efficient evaluation of evidential strength required in criminal cases. The problems arise from stochastic phenomena such as "dropout" (absence of an allele in the profile that is present in the underlying DNA) and experimental artefacts such as "stutter" that can generate peaks of ambiguous allelic status. Currently in the UK, evidential strength evaluation uses an approach in which the complex signals in the DNA profiles are interpreted in a semi-manual fashion by trained experts aided by a set of guidelines, but also relying substantially on professional judgment. We introduce a statistical model to calculate likelihood ratios for evaluating DNA evidence arising from multiple known and unknown contributors that allows for such stochastic phenomena by incorporating peak heights. Efficient use of peak heights allows for more crime scene profiles to be reported to courts than is currently possible. The model parameters are estimated from experimental data incorporating multiple sources of variability in the profiling system. We report and analyse experimental results from the SGMPlus system, run at 28 amplification cycles with no enhancements, currently used in the UK. Our methods are readily adapted to other DNA profiling systems provided that the experimental data for the parameter estimation is available.

Consideration of the probative value of single donor 15-plex STR profiles in UK populations and its presentation in UK courts.

The adoption of new 15 locus STR multiplex systems into UK forensic science would be facilitated by agreed guidelines for reporting the strength of DNA evidence using likelihood ratios. To facilitate such an agreement, we present an analysis of previously published UK allele frequencies for white Caucasian, Afro-Caribbean and Indo-Pakistani populations and investigate their effect on likelihood ratios for single donor profiles. We consider the implication of the five additional loci and suggest a procedure for reporting likelihood ratios for 15-plex STR profiles.

Calculating likelihood ratios for a mixed DNA profile when a contribution from a genetic relative of a suspect is proposed.

This technical note describes a practical method for evaluating evidence in the case of a two person conditioned DNA mixture where the defence proposition is that the unknown contributor is genetically related to the suspect. A conditioned mixture is one where the presence of DNA from one of two individuals is accepted by both prosecution and defence. A typical example would be a vaginal swab in an alleged rape case, where the presence of the complainant's DNA would be expected and samples have been taken from the complainant and a suspect. Much has been written about the interpretation of such mixtures and the calculation of the conditional genotype probabilities that must be carried out. In general, such treatments assume that the unknown contributor, under the defence proposition, is unrelated to the known individuals. In this paper, we consider the case where the defence proposition is that the unknown contributor is genetically related to the suspect. We describe a method, incorporating a flow chart and reference tables that facilitate manual calculations of the likelihood ratio for several postulated genetic relationships.

The low-template-DNA (stochastic) threshold--its determination relative to risk analysis for national DNA databases.

Although the low-template or stochastic threshold is in widespread use and is typically set to 150-200 rfu peak height, there has been no consideration on its determination and meaning. In this paper we propose a definition that is based upon the specific risk of wrongful designation of a heterozygous genotype as a homozygote which could lead to a false exclusion. Conversely, it is possible that a homozygote {a,a} could be designated as {a,F} where 'F' is a 'wild card', and this could lead to increased risk of false inclusion. To determine these risk levels, we analysed an experimental dataset that exhibited extreme drop-out using logistic regression. The derived probabilities are employed in a graphical model to determine the relative risks of wrongful designations that may cause false inclusions and exclusions. The methods described in this paper provide a preliminary solution of risk evaluation for any DNA process that employs a stochastic threshold.

Computation of likelihood ratios in fingerprint identification for configurations of any number of minutiae.

Recent court challenges have highlighted the need for statistical research on fingerprint identification. This paper proposes a model for computing likelihood ratios (LRs) to assess the evidential value of comparisons with any number of minutiae. The model considers minutiae type, direction and relative spatial relationships. It expands on previous work on three minutiae by adopting a spatial modeling using radial triangulation and a probabilistic distortion model for assessing the numerator of the LR. The model has been tested on a sample of 686 ulnar loops and 204 arches. Features vectors used for statistical analysis have been obtained following a preprocessing step based on Gabor filtering and image processing to extract minutiae data. The metric used to assess similarity between two feature vectors is based on an Euclidean distance measure. Tippett plots and rates of misleading evidence have been used as performance indicators of the model. The model has shown encouraging behavior with low rates of misleading evidence and a LR power of the model increasing significantly with the number of minutiae. The LRs that it provides are highly indicative of identity of source on a significant proportion of cases, even when considering configurations with few minutiae. In contrast with previous research, the model, in addition to minutia type and direction, incorporates spatial relationships of minutiae without introducing probabilistic independence assumptions. The model also accounts for finger distortion.

Computation of likelihood ratios in fingerprint identification for configurations of three minutiae.

Recent challenges to fingerprint evidence have brought forward the need for peer-reviewed scientific publications to support the evidential value assessment of fingerprint. This paper proposes some research directions to gather statistical knowledge of the within-source and between-sources variability of configurations of three minutiae on fingermarks and fingerprints. This paper proposes the use of the likelihood ratio (LR) approach to assess the value of fingerprint evidence. The model explores the statistical contribution of configurations of three minutiae using Tippett plots and related measures to assess the quality of the system. Features vectors used for statistical analysis have been obtained following a preprocessing step based on Gabor filtering and image processing to extract minutia position, type, and direction. Spatial relationships have been coded using Delaunay triangulation. The metric, used to assess similarity between two feature vectors is based on an Euclidean distance measure. The within-source variability has been estimated using a sample of 216 fingerprints from four fingers (two donors). Between-sources variability takes advantage of a database of 818 ulnar loops from randomly selected males. The results show that the data-driven approach adopted here is robust. The magnitude of LRs obtained under the prosecution and defense propositions stresses upon the major evidential contribution that small portions of fingermark, containing three minutiae, can provide regardless of its position on the general pattern.