A large-scale dataset of single and mixed-source short tandem repeat profiles to inform human identification strategies: PROVEDIt

À la carte: how mosquitoes choose their blood meals

Mosquitoes are important vectors for human diseases, transmitting pathogens that cause a range of parasitic and viral infections. Mosquito blood-feeding is heterogeneous, meaning that some human hosts are at higher risk of receiving bites than others, and this heterogeneity is multifactorial. Mosquitoes integrate specific cues to locate their hosts, and mosquito attraction differs considerably between individual human hosts. Heterogeneous mosquito biting results from variations in both host attractiveness and availability and can impact transmission of vector-borne diseases. However, the extent and drivers of this heterogeneity and its importance for pathogen transmission remain incompletely understood. Here, we review methods and recent data describing human characteristics that affect host-seeking behavior and host preferences of mosquito disease vectors, and the implications for vector-borne disease transmission.

Extending the discussion on inconsistency in forensic decisions and results

The subject of inter- and intra-laboratory inconsistency was recently raised in a commentary by Itiel Dror. We re-visit an inter-laboratory trial, with which some of the authors of this current discussion were associated, to diagnose the causes of any differences in the likelihood ratios (LRs) assigned using probabilistic genotyping software. Some of the variation was due to different decisions that would be made on a case-by-case basis, some due to laboratory policy and would hence differ between laboratories, and the final and smallest part was the run-to-run difference caused by the Monte Carlo aspect of the software used. However, the net variation in LRs was considerable. We believe that most laboratories will self-diagnose the cause of their difference from the majority answer and in some, but not all instances will take corrective action. An inter-laboratory exercise consisting of raw data files for relatively straightforward mixtures, such as two mixtures of three or four persons, would allow laboratories to calibrate their procedures and findings.

Machine learning applications in forensic DNA profiling: A critical review

Machine learning (ML) is a range of powerful computational algorithms capable of generating predictive models via intelligent autonomous analysis of relatively large and often unstructured data. ML has become an integral part of our daily lives with a plethora of applications, including web, business, automotive industry, clinical diagnostics, scientific research, and more recently, forensic science. In the field of forensic DNA, the manual analysis of complex data can be challenging, time-consuming, and error-prone. The integration of novel ML-based methods may aid in streamlining this process while maintaining the high accuracy and reproducibility required for forensic tools. Due to the relative novelty of such applications, the forensic community is largely unaware of ML capabilities and limitations. Furthermore, computer science and ML professionals are often unfamiliar with the forensic science field and its specific requirements. This manuscript offers a brief introduction to the capabilities of machine learning methods and their applications in the context of forensic DNA analysis and offers a critical review of the current literature in this rapidly developing field.

bistro: An R package for vector bloodmeal identification by short tandem repeat overlap

Measuring vector-human contact in a natural setting can inform precise targeting of interventions to interrupt transmission of vector-borne diseases. One approach is to directly match human DNA in vector bloodmeals to the individuals who were bitten using genotype panels of discriminative short tandem repeats (STRs). Existing methods for matching STR profiles in bloodmeals to the people bitten preclude the ability to match most incomplete profiles and multi-source bloodmeals to bitten individuals.We developed bistro, an R package that implements 3 preexisting STR matching methods as well as the package's namesake, bistro, a new algorithm described here. bistro employs forensic analysis methods to calculate likelihood ratios and match human STR profiles in bloodmeals to people using a dynamic threshold. We evaluated the algorithm's accuracy and compared it to existing matching approaches using a publicly-available panel of 188 single-source and 100 multi-source samples containing DNA from 50 known human sources. Then we applied it to match 777 newly field-collected mosquito bloodmeals to a database of 645 people.The R package implements four STR matching algorithms in user-friendly functions with clear documentation. bistro correctly matched 99% (187/188) of profiles in single-source samples, and 62% (224/359) of profiles from multi-source samples, resulting in a sensitivity of 0.75 (vs < 0.51 for other algorithms). The specificity of bistro was 0.9998 (vs. 1 for other algorithms). Furthermore, bistro identified 79% (720/906) of all possible matches for field-derived mosquitoes, yielding 1.4x more matches than existing algorithms.bistro identifies more correct bloodmeal-human matches than existing approaches, enabling more accurate and robust analyses of vector-human contact in natural settings. The bistro R package and corresponding documentation allow for straightforward uptake of this algorithm by others.

Single cell genomics applications in forensic science: Current state and future directions

Standard methods of mixture analysis involve subjecting a dried crime scene sample to a "bulk" DNA extraction method such that the resulting isolate compromises a homogenized DNA mixture from the individual donors. If, however, instead of bulk DNA extraction, a sufficient number of individual cells from the mixed stain are subsampled prior to genetic analysis then it should be possible to recover highly probative single source, non-mixed scDNA profiles from each of the donors. This approach can detect low DNA level minor donors to a mixture that otherwise would not be identified using standard methods and can also resolve rare mixtures comprising first degree relatives and thereby also prevent the false inclusion of non-donor relatives. This literature landscape review and associated commentary reports on the history and increasing interest in current and potential future applications of scDNA in forensic genomics, and critically evaluates opportunities and impediments to further progress.

Effect existence of aging on stutter ratio evaluated via Bayesian inference

The stochastic behavior of the stutter ratio (SR) in capillary electrophoresis-based DNA typing is currently described and predicted using statistical models in forensic genetics. Clarifying this behavior can help obtain more objective and robust evidence to the court in terms of mixture interpretation. This study aimed to investigate the effect existence of aging on SR via a Bayesian framework. Nail scrapings and clippings were collected from 68 healthy individuals with informed consent. Samples were classified by age-class: young group (0-16 years; n = 36) and older-adult group (>61 years; n = 32). Then, they were compared in terms of their SRs for each simple repeat locus included in GlobalFiler Kit. Bayesian modeling was performed with lognormal distribution model, which implemented multiple linear regression, allele and age-class as explanatory variables. For all simple repeat loci, the median of the posterior distribution of the age-class parameter was a positive value. For CSF1PO and D7S820, the 95% credible interval of the posterior distribution did not include 0. Our data suggested that aging slightly increases the SR. These findings might help elucidate the stochastic behavior of SR.

Separation mixed semen of two individuals using magnetic beads coupled ABH blood group antibody

In sexual assault cases, one of the most common samples collected is a mixed semen stain, which is often found on the vagina, female underwear, or bed sheets. However, it is usually difficult to identify the perpetrator based on this sample alone. One technique that has been developed to address this issue is magnetic bead-based separation. This method involves using modified magnetic microspheres to capture and enrich specific target cells, in this case, sperm cells. In this study, we utilized magnetic beads coupled with ABH blood group antibody to isolate sperm cells from an individual of a single ABO blood type. Subsequently, polymerase chain reaction amplification and capillary electrophoresis were employed to perform the genotyping the short tandem repeat (STR) loci. This approach allows for the identification of different individuals in a mixed seminal stain sample from two individuals, by first separating sperm cells based on ABH antigen differences and subsequently utilizing autosomal STR typing on the enriched single blood group cells.

bistro: An R package for vector bloodmeal identification by short tandem repeat overlap

1. Measuring vector-human contact in a natural setting can inform precise targeting of interventions to interrupt transmission of vector-borne diseases. One approach is to directly match human DNA in vector bloodmeals to the individuals who were bitten using genotype panels of discriminative short tandem repeats (STRs). Existing methods for matching STR profiles in bloodmeals to the people bitten preclude the ability to match most incomplete profiles and multi-source bloodmeals to bitten individuals. 2. We developed bistro, an R package that implements 3 preexisting STR matching methods as well as the package's namesake, bistro, a new algorithm described here. bistro employs forensic analysis methods to calculate likelihood ratios and match human STR profiles in bloodmeals to people using a dynamic threshold. We evaluated the algorithm's accuracy and compared it to existing matching approaches using a publicly-available panel of 188 single-source and 100 multi-source samples containing DNA from 50 known human sources. Then we applied it to match 777 newly field-collected mosquito bloodmeals to a database of 645 people. 3. The R package implements four STR matching algorithms in user-friendly functions with clear documentation. bistro correctly matched 99% (184/185) of profiles in single-source samples, and 63% (225/359) of profiles from multi-source samples, resulting in a sensitivity of 0.75 (vs < 0.51 for other algorithms). The specificity of bistro was 0.9998 (vs. 1 for other algorithms). Furthermore, bistro identified 80% (729/909) of all possible matches for field-derived mosquitoes, yielding 1.4x more matches than existing algorithms. 4. bistro identifies more correct bloodmeal-human matches than existing approaches, enabling more accurate and robust analyses of vector-human contact in natural settings. The bistro R package and corresponding documentation allow for straightforward uptake of this algorithm by others.

Addressing uncertain assumptions in DNA evidence evaluation

Evidential value of DNA mixtures is typically expressed by a likelihood ratio. However, selecting appropriate propositions can be contentious, because assumptions may need to be made around, for example, the contribution of a complainant's profile, or relatedness between contributors. A choice made one way or another disregards any uncertainty that may be present about such an assumption. To address this, a complex proposition that considers multiple sub-propositions with different assumptions may be more appropriate. While the use of complex propositions has been advocated in the literature, the uptake in casework has been limited. We provide a mathematical framework for evaluating DNA evidence given complex propositions and discuss its implementation in the DBLR™ software. The software simultaneously handles multiple mixed samples, reference profiles and relationships as described by a pedigree, which unlocks a variety of applications. We provide several examples to illustrate how complex propositions can efficiently evaluate DNA evidence. The addition of this feature to DBLR™ provides a tool to approach the long-accepted, but often impractical suggestion that propositions should be exhaustive within a case context.

STRavinsky STR database and PGTailor PGT tool demonstrate superiority of CHM13-T2T over hg38 and hg19 for STR-based applications

Short-Tandem-Repeats (STRs) have long been studied for possible roles in biological phenomena, and are utilized in multiple applications such as forensics, evolutionary studies and pre-implantation-genetic-testing (PGT). The two reference genomes most used by clinicians and researchers are GRCh37/hg19 and GRCh38/hg38, both constructed using mainly short-read-sequencing (SRS) in which all-STR-containing-reads cannot be assembled to the reference genome. With the introduction of long-read-sequencing (LRS) methods and the generation of the CHM13 reference genome, also known as T2T, many previously unmapped STRs were finally localized within the human genome. We generated STRavinsky, a compact STR database for three reference genomes, including T2T. We proceeded to demonstrate the advantages of T2T over hg19 and hg38, identifying nearly double the number of STRs throughout all chromosomes. Through STRavinsky, providing a resolution down to a specific genomic coordinate, we demonstrated extreme propensity of TGGAA repeats in p arms of acrocentric chromosomes, substantially corroborating early molecular studies suggesting a possible role in formation of Robertsonian translocations. Moreover, we delineated unique propensity of TGGAA repeats specifically in chromosome 16q11.2 and in 9q12. Finally, we harness the superior capabilities of T2T and STRavinsky to generate PGTailor, a novel web application dramatically facilitating design of STR-based PGT tests in mere minutes.

Variational inference accelerates accurate DNA mixture deconvolution

We investigate a class of DNA mixture deconvolution algorithms based on variational inference, and we show that this can significantly reduce computational runtimes with little or no effect on the accuracy and precision of the result. In particular, we consider Stein Variational Gradient Descent (SVGD) and Variational Inference (VI) with an evidence lower-bound objective. Both provide alternatives to the commonly used Markov-Chain Monte-Carlo methods for estimating the model posterior in Bayesian probabilistic genotyping. We demonstrate that both SVGD and VI significantly reduce computational costs over the current state of the art. Importantly, VI does so without sacrificing precision or accuracy, presenting an overall improvement over previously published methods.

Analysis of the Hamiltonian Monte Carlo genotyping algorithm on PROVEDIt mixtures including a novel precision benchmark

We provide an internal validation study of a recently published precise DNA mixture algorithm based on Hamiltonian Monte Carlo sampling (Susik et al., 2022). We provide results for all 428 mixtures analysed by Riman et al. (2021) and compare the results with two state-of-the-art software products: STRmix™  v2.6 and Euroformix v3.4.0. The comparison shows that the Hamiltonian Monte Carlo method provides reliable values of likelihood ratios (LRs) close to the other methods. We further propose a novel large-scale precision benchmark and quantify the precision of the Hamiltonian Monte Carlo method, indicating its improvements over existing solutions. Finally, we analyse the influence of the factors discussed by Buckleton et al. (2022).

Hamiltonian Monte Carlo with strict convergence criteria reduces run-to-run variability in forensic DNA mixture deconvolution

MOTIVATION

Analysing mixed DNA profiles is a common task in forensic genetics. Due to the complexity of the data, such analysis is often performed using Markov Chain Monte Carlo (MCMC)-based genotyping algorithms. These trade off precision against execution time. When default settings (including default chain lengths) are used, as large as a 10-fold changes in inferred log-likelihood ratios (LR) are observed when the software is run twice on the same case. So far, this uncertainty has been attributed to the stochasticity of MCMC algorithms. Since LRs translate directly to strength of the evidence in a criminal trial, forensic laboratories desire LR with small run-to-run variability.

RESULTS

We present the use of a Hamiltonian Monte Carlo (HMC) algorithm that reduces run-to-run variability in forensic DNA mixture deconvolution by around an order of magnitude without increased runtime. We achieve this by enforcing strict convergence criteria. We show that the choice of convergence metric strongly influences precision. We validate our method by reproducing previously published results for benchmark DNA mixtures (MIX05, MIX13, and ProvedIt). We also present a complete software implementation of our algorithm that is able to leverage GPU acceleration for the inference process. In the benchmark mixtures, on consumer-grade hardware, the runtime is less than 7 min for 3 contributors, less than 35 min for 4 contributors, and less than an hour for 5 contributors with one known contributor.

A tool for simulating single source and mixed DNA profiles

Simulation studies play an important role in the study of probabilistic genotyping systems, as a low cost and fast alternative to in vitro studies. With ongoing calls for further study of the behaviour of probabilistic genotyping systems, there is a continuous need for such studies. In most cases, researchers use simplified models, for example ignoring complexities such as peak height variability due to lack of availability of advanced tools. We fill this void and describe a tool that can simulate DNA profiles in silico for the validation and investigation of probabilistic genotyping software. Contributor genotypes are simulated by randomly sampling alleles from selected allele frequencies. Some or all contributors may be related to a pedigree and the genotypes of non-founders are obtained by random gene dropping. The number of contributors per profile, and ranges for parameters such as DNA template amount and degradation parameters can be configured. Peak height variability is modelled using a lognormal distribution or a gamma distribution. Profile behaviour of simulated profiles is shown to be broadly similar to laboratory generated profiles though the latter shows more variation. Simulation studies do not remove the need for experimental data. The tool has been made available as an R-package named simDNAmixtures.

Exploring likelihood ratios assigned for siblings of the true mixture contributor as an alternate contributor

Relatives tend to have more DNA in common than unrelated people. The closer the biological relationship, the higher the chance of alleles being identical by descent between the individuals. Therefore, when considering a mixed DNA profile, close relatives of the true contributor may not always be excluded as a possible contributor to a mixture due to allele sharing. In these situations, it might be more appropriate under the alternate proposition to consider that the DNA could have originated from a relative of the person of interest rather than an unrelated individual. The probabilistic genotyping software STRmix™ automatically provides LRs considering close biological relatives as alternate sources of the DNA. In this paper, we investigate the support for siblings of the true contributor to a mixture (who are not present in the mixture themselves). We interpret the mixtures and assign LRs using STRmix™ and investigate whether the resulting LRs could be used to indicate whether the true contributor could be a sibling of the POI. Most siblings will have one or more alleles that are not observed in the mixture profile. Support for siblings to have contributed can only occur when allelic dropout is a possibility at the loci where the siblings have alleles that are not observed in the profile. In these data, that was only observed in components with assigned template of 588 rfu or less.

TAWSEEM: A Deep-Learning-Based Tool for Estimating the Number of Unknown Contributors in DNA Profiling

DNA profiling involves the analysis of sequences of an individual or mixed DNA profiles to identify the persons that these profiles belong to. A critically important application of DNA profiling is in forensic science to identify criminals by finding a match between their blood samples and the DNA profile found on the crime scene. Other applications include paternity tests, disaster victim identification, missing person investigations, and mapping genetic diseases. A crucial task in DNA profiling is the determination of the number of contributors in a DNA mixture profile, which is challenging due to issues that include allele dropout, stutter, blobs, and noise in DNA profiles; these issues negatively affect the estimation accuracy and the computational complexity. Machine-learning-based methods have been applied for estimating the number of unknowns; however, there is limited work in this area and many more efforts are required to develop robust models and their training on large and diverse datasets. In this paper, we propose and develop a software tool called TAWSEEM that employs a multilayer perceptron (MLP) neural network deep learning model for estimating the number of unknown contributors in DNA mixture profiles using PROVEDIt, the largest publicly available dataset. We investigate the performance of our developed deep learning model using four performance metrics, namely accuracy, F1-score, recall, and precision. The novelty of our tool is evident in the fact that it provides the highest accuracy (97%) compared to any existing work on the most diverse dataset (in terms of the profiles, loci, multiplexes, etc.). We also provide a detailed background on the DNA profiling and literature review, and a detailed account of the deep learning tool development and the performance investigation of the deep learning method.

Using a multi-head, convolutional neural network with data augmentation to improve electropherogram classification performance

DNA profiles are generated in forensic biology laboratories around the world. It is possible that these profiles are assessed by two independent people in order for the profiles to be 'read'. Recent work has been carried out to develop a neural network model to classify fluorescence in a DNA profile electropherogram and potentially replace one, or both human readers. The ability to use neural networks for this function has been programmed into the software FaSTR™ DNA, which has been validated for use in at least one laboratory in Australia. The work that previously developed a neural network system had a number of limitations, specifically it was computer intensive, did not make the best use of available data, and consequently the performance of this model was sub-optimal in some conditions (particularly for low-intensity peaks). In the current work a new neural network model is developed that makes various improvements on the old model, by using convolutional layers, a multi-head architecture and data augmentation. Results indicate that an improved performance can be expected for low-intensity profiles.

A mixed DNA profile controversy revisited

Semaan et al. (J Forensic Res, 2020, 11, 453) discuss a mock case "where eight different individuals [P₁ through P₈ ] could not be excluded in a mixed DNA analysis. Even though … expert DNA mixture analysis software was used." Two of these are the true donors. The LRs reported are incorrect due to the incorrect entry of propositions into LRmix Studio. This forced the software to account for most of the alleles as drop-in, resulting in LRs 60-70 orders of magnitude larger than expected. P₁ , P₂ , P₄ , P₅ , and P₈ can be manually excluded using peak heights. This has relevance when using LRmix which does not use peak heights. We extend the work using the same two reference genotypes who were the true contributors as Semaan et al. (J Forensic Res, 2020, 11, 453). We simulate three two-donor mixtures with peak heights using these two genotypes and analyze using STRmix™. For the simulated 1:1 mixture, one of the non-donors' LRs supported him being a contributor when no conditioning was used. When considered in combination with any other potential donors (i.e., with conditioning), this non-donor was correctly eliminated. For the 3:1 mixture, all results correctly supported that the non-donors were not contributors. The low-template 4:1 mixture LRs with no conditioning showed support for all eight profiles as donors. However, the results from pair-wise conditioning showed that only the two ground truth donors had LRs supporting that they were contributors to the mixture. We recommend the use of peak heights and conditioning profiles, as this allows better sensitivity and specificity even when the persons share many alleles.

Development and validation of Kongoh ver. 3.0.1: Open-source software for DNA mixture interpretation in the GlobalFiler system based on a quantitative continuous model

Probabilistic genotyping software based on continuous models is effective for interpreting DNA profiles derived from DNA mixtures and small DNA samples. In this study, we updated our previously developed Kongoh software (to ver. 3.0.1) to interpret DNA profiles typed using the GlobalFiler™ PCR Amplification Kit. Recently, highly sensitive typing systems such as the GlobalFiler system have facilitated the detection of forward, double-back, and minus 2-nt stutters; therefore, we implemented statistical models for these stutters in Kongoh. In addition, we validated the new version of Kongoh using 2-4-person mixtures and DNA profiles with degradation in the GlobalFiler system. The likelihood ratios (LRs) for true contributors and non-contributors were well separated as the information increased (i.e., larger peak height and fewer contributors), and these LRs tended to neutrality as the information decreased. These trends were observed even in profiles with DNA degradation. The LR values were highly reproducible, and the accuracy of the calculation was also confirmed. Therefore, Kongoh ver. 3.0.1 is useful for interpreting DNA mixtures and degraded DNA samples in the GlobalFiler system.

A comparison of likelihood ratios obtained from EuroForMix and STRmix™

Likelihood ratios (LR) differences between the probabilistic genotyping software EuroForMix and STRmix™ are examined. After considering differences in the allele probabilities, the LRs from both software for an unambiguous single-source profile were identical (four significant figures). LRs from both software for an unambiguous single-source profile with alleles previously unseen in the allele frequency database (rare alleles) were the same (three significant figures) for θ = 0.01. Due to differences in the minimum allele frequencies, the LRs differed by three orders of magnitude when θ = 0. For both software, the LRs for a single-source dilution series decreased as the input amount decreased. The LRs from both software were within an order of magnitude for known contributors. The largest difference was where the target input amount was 0.0156 ng: The LR_EuroForMix was 2.1 × 10²⁵ and the LR_STRmix was 8.0 × 10²⁴ . Both software show similar LR behavior with respect to mixture ratio. For two person mixtures the LR increases for both the major and the minor as the ratio moves away from 1:1. The LR for the major stabilizes at about 3:1 whereas the LR for the minor reaches its maximum at about 3:1 and then declines. Greater differences in LR were observed between EuroForMix and STRmix™ for mixtures. One-hundred and twenty-nine mixtures from the PROVEDIt dataset were compared. LRs for 84% of the comparisons for known contributors without rare alleles were within two orders of magnitude. Five divergent results were investigated, and a manual intervention approach was applied where appropriate.

Examining performance and likelihood ratios for two likelihood ratio systems using the PROVEDIt dataset

A likelihood ratio (LR) system is defined as the entire pipeline of the measurement and interpretation processes where probabilistic genotyping software (PGS) is a piece of the whole LR system. To gain understanding on how two LR systems perform, a total of 154 two-person, 147 three-person, and 127 four-person mixture profiles of varying DNA quality, DNA quantity, and mixture ratios were obtained from the filtered (.CSV) files of the GlobalFiler 29 cycles 15s PROVEDIt dataset and deconvolved in two independently developed fully continuous programs, STRmix v2.6 and EuroForMix v2.1.0. Various parameters were set in each software and LR computations obtained from the two software were based on same/fixed EPG features, same pair of propositions, number of contributors, theta, and population allele frequencies. The ability of each LR system to discriminate between contributor (H1-true) and non-contributor (H2-true) scenarios was evaluated qualitatively and quantitatively. Differences in the numeric LR values and their corresponding verbal classifications between the two LR systems were compared. The magnitude of the differences in the assigned LRs and the potential explanations for the observed differences greater than or equal to 3 on the log10 scale were described. Cases of LR < 1 for H1-true tests and LR > 1 for H2-true tests were also discussed. Our intent is to demonstrate the value of using a publicly available ground truth known mixture dataset to assess discrimination performance of any LR system and show the steps used to understand similarities and differences between different LR systems. We share our observations with the forensic community and describe how examining more than one PGS with similar discrimination power can be beneficial, help analysts compare interpretation especially with low-template profiles or minor contributor cases, and be a potential additional diagnostic check even if software in use does contain certain diagnostic statistics as part of the output.

The a posteriori probability of the number of contributors when conditioned on an assumed contributor

Forensic DNA signal is notoriously challenging to assess, requiring computational tools to support its interpretation. Over-expressions of stutter, allele drop-out, allele drop-in, degradation, differential degradation, and the like, make forensic DNA profiles too complicated to evaluate by manual methods. In response, computational tools that make point estimates on the Number of Contributors (NOC) to a sample have been developed, as have Bayesian methods that evaluate an A Posteriori Probability (APP) distribution on the NOC. In cases where an overly narrow NOC range is assumed, the downstream strength of evidence may be incomplete insofar as the evidence is evaluated with an inadequate set of propositions. In the current paper, we extend previous work on NOCIt, a Bayesian method that determines an APP on the NOC given an electropherogram, by reporting on an implementation where the user can add assumed contributors. NOCIt is a continuous system that incorporates models of peak height (including degradation and differential degradation), forward and reverse stutter, noise, and allelic drop-out, while being cognizant of allele frequencies in a reference population. When conditioned on a known contributor, we found that the mode of the APP distribution can shift to one greater when compared with the circumstance where no known contributor is assumed, and that occurred most often when the assumed contributor was the minor constituent to the mixture. In a development of a result of Slooten and Caliebe (FSI:G, 2018) that, under suitable assumptions, establishes the NOC can be treated as a nuisance variable in the computation of a likelihood ratio between the prosecution and defense hypotheses, we show that this computation must not only use coincident models, but also coincident contextual information. The results reported here, therefore, illustrate the power of modern probabilistic systems to assess full weights-of-evidence, and to provide information on reasonable NOC ranges across multiple contexts.

A series of developmental validation tests for Number of Contributors platforms: Exemplars using NOCIt and a neural network

Complex DNA mixtures are challenging to interpret and require computational tools that aid in that interpretation. Recently, several computational methods that estimate the number of contributors (NOC) to a sample have been developed. Unlike analogous tools that interpret profiles and report LRs, NOC tools vary widely in their operational principle where some are Bayesian and others are machine learning tools. Conjunctionally, NOC tools may return a single n estimate, or a distribution on n. This vast array of constructs, coupled with a gap in standardized methods by which to validate NOC systems, warrants an exploration into the measures by which differing NOC systems might be tested for operations. In the current paper, we use two exemplar NOC systems: a probabilistic system named NOCIt, which renders an a posteriori probability (APP) distribution on the number of contributors given an electropherogram and an artificial neural network (ANN). NOCIt is a continuous Bayesian inference system incorporating models of peak height, degradation, differential degradation, forward and reverse stutter, noise and allelic drop-out while considering allele frequencies in a reference population. The ANN is also a continuous method, taking all the same features (barring degradation) into account. Unlike its Bayesian counterpart, it demands substantively more data to parameterize, requiring synthetic data. We explore each system's performance by conducting tests on 214 PROVEDIt mixtures where the limit of detection was 1-copy of DNA. We found that after a lengthy training period of approximately 24 h, the ANN's evaluation process was very fast and perfectly repeatable. In contrast, NOCIt only took a few minutes to train but took tens of minutes to complete each sample and was less repeatable. In addition, it rendered a probability distribution that was more sensitive and specific, affording a reasonable method by which to report all reasonable n that explain the evidence for a given sample. Whatever the method, by acknowledging the inherent differences between NOC systems, we demonstrate that validation constructs will necessarily be guided by the needs of the forensic domain and be dependent upon whether the laboratory seeks to assign a single n or range of n.

Evaluating DNA evidence possibly involving multiple (mixed) samples, common donors and related contributors

Forensic DNA profiling is used in various circumstances to evaluate support for two competing propositions with the assignment of a likelihood ratio. Many software implementations exist that tackle a range of inference problems spanning identification and relationship testing. We propose a flexible likelihood ratio framework that caters to inference problems in forensic genetics. The framework allows for investigation of the degree of support for the contribution of multiple persons to multiple samples allowing for persons to be related according to a pedigree, including inbred relationships. We explain how a number of routine as well as more complex problems can be treated within this framework.

Comparing multiple POI to DNA mixtures

In casework, laboratories may be asked to compare DNA mixtures to multiple persons of interest (POI). Guidelines on forensic DNA mixture interpretation recommend that analysts consider several pairs of propositions; however, it is unclear if several likelihood ratios (LRs) per person should be reported or not. The propositions communicated to the court should not depend on the value of the LR. As such, we suggest that the propositions should be functionally exhaustive. This implies that all propositions with a non-zero prior probability need to be considered, at least initially. Those that have a significant posterior probability need to be used in the final evaluation. Using standard probability theory we combine various propositions so that collectively they are exhaustive. This involves a prior probability that the sub-proposition is true, given that the primary proposition is true. Imagine a case in which there are two possible donors: i and j. We focus our analysis first on donor i so that the primary proposition is that i is one of the sources of the DNA. In this example, given that i is a donor, we would further consider that j is either a donor or not. In practice, the prior weights for these sub-propositions may be difficult to assign. However, the LR is often linearly related to these priors and its behaviour is predictable. We also believe that these priors are unavoidable and are hidden in alternative methods. We term the likelihood ratio formed from these context-exhaustive propositions LR_i/i¯. LR_i/i¯ is trialed in a set of two- and three-person mixtures. For two-person mixtures, LR_i/i¯ is often well approximated by LR_ij/ja, where the subscript ij describes the proposition that i and j are the donors and ja describes the proposition that j and an alternate, unknown individual (a), who is unrelated to both i and j, are the donors. For three-person mixtures, LR_i/i¯ is often well approximated by LR_ijk/jka where the subscript ijk describes the proposition that i, j, and k are the donors and jka describes the proposition that j, k, and an unknown, unrelated (to i, j, and k) individual (a) are the donors. In our simulations, LR_ij/ja had fewer inclusionary LRs for non-contributors than the unconditioned LR (LR_ia/aa).

Validation of a top-down DNA profile analysis for database searching using a fully continuous probabilistic genotyping model

Slooten described a method of targeting major contributors in mixed DNA profiles and comparing them to individuals on a DNA database. The method worked by taking incrementally more peak information from the profile (based on the peak contribution), and using a semi-continuous model, calculating likelihood ratios for the comparison to database individuals. We describe the performance of this "top down approach" to profile interpretation within probabilistic genotyping software employing a fully continuous model. We interpret both complex constructed profiles where ground truth is known and casework profiles from non-suspect crimes. The interpretation of constructed four- and five- person mixtures demonstrated good discrimination power between contributors and non-contributors to the mixtures. Not all known contributors linked, and this is expected, particularly for minor contributors of DNA to the profile, or when the DNA from contributors was in relatively equal contributions. This finding was also reported by Slooten for the semi-continuous application of the approach. The maximum observed LR was shown to not exceed the LR obtained after a standard interpretation approach outside of that expected due to Monte Carlo variation. The interpretation of 91 complex profiles from no-suspect casework demonstrated that approximately 75% of profiles returned a link to someone on a database of known individuals. With a yearly average of 110 no-suspect cases that fall into this too-complex category at Forensic Science SA, the top down analysis, if applied to all such profiles, would represent an increase of 83 links per year of investigative information that could be provided to investigators.

Towards developing forensically relevant single-cell pipelines by incorporating direct-to-PCR extraction: compatibility, signal quality, and allele detection

Current analysis of forensic DNA stains relies on the probabilistic interpretation of bulk-processed samples that represent mixed profiles consisting of an unknown number of potentially partial representations of each contributor. Single-cell methods, in contrast, offer a solution to the forensic DNA mixture problem by incorporating a step that separates cells before extraction. A forensically relevant single-cell pipeline relies on efficient direct-to-PCR extractions that are compatible with standard downstream forensic reagents. Here we demonstrate the feasibility of implementing single-cell pipelines into the forensic process by exploring four metrics of electropherogram (EPG) signal quality-i.e., allele detection rates, peak heights, peak height ratios, and peak height balance across low- to high-molecular-weight short tandem repeat (STR) markers-obtained with four direct-to-PCR extraction treatments and a common post-PCR laboratory procedure. Each treatment was used to extract DNA from 102 single buccal cells, whereupon the amplification reagents were immediately added to the tube and the DNA was amplified/injected using post-PCR conditions known to elicit a limit of detection (LoD) of one DNA molecule. The results show that most cells, regardless of extraction treatment, rendered EPGs with at least a 50% true positive allele detection rate and that allele drop-out was not cell independent. Statistical tests demonstrated that extraction treatments significantly impacted all metrics of EPG quality, where the Arcturus® PicoPure™ extraction method resulted in the lowest median allele drop-out rate, highest median average peak height, highest median average peak height ratio, and least negative median values of EPG sloping for GlobalFiler™ STR loci amplified at half volume. We, therefore, conclude the feasibility of implementing single-cell pipelines for casework purposes and demonstrate that inferential systems assuming cell independence will not be appropriate in the probabilistic interpretation of a collection of single-cell EPGs.

Estimating the number of contributors to a DNA profile using decision trees

The interpretation of DNA profiles typically starts with an assessment of the number of contributors. In the last two decades, several methods have been proposed to assist with this assessment. We describe a relatively simple method using decision trees, that is fast to run and fully transparent to a forensic analyst. We use mixtures from the publicly available PROVEDIt dataset to demonstrate the performance of the method. We show that the performance of the method crucially depends on the performance of filters for stutter and other artefacts. We compare the performance of the decision tree method with other published methods for the same dataset.

Interpol review of forensic biology and forensic DNA typing 2016-2019

This review paper covers the forensic-relevant literature in biological sciences from 2016 to 2019 as a part of the 19th Interpol International Forensic Science Managers Symposium. The review papers are also available at the Interpol website at: https://www.interpol.int/content/download/14458/file/Interpol%20Review%20Papers%202019.pdf.

Statistical modeling of STR capillary electrophoresis signal

BACKGROUND

In order to isolate an individual's genotype from a sample of biological material, most laboratories use PCR and Capillary Electrophoresis (CE) to construct a genetic profile based on polymorphic loci known as Short Tandem Repeats (STRs). The resulting profile consists of CE signal which contains information about the length and number of STR units amplified. For samples collected from the environment, interpretation of the signal can be challenging given that information regarding the quality and quantity of the DNA is often limited. The signal can be further compounded by the presence of noise and PCR artifacts such as stutter which can mask or mimic biological alleles. Because manual interpretation methods cannot comprehensively account for such nuances, it would be valuable to develop a signal model that can effectively characterize the various components of STR signal independent of a priori knowledge of the quantity or quality of DNA.

RESULTS

First, we seek to mathematically characterize the quality of the profile by measuring changes in the signal with respect to amplicon size. Next, we examine the noise, allele, and stutter components of the signal and develop distinct models for each. Using cross-validation and model selection, we identify a model that can be effectively utilized for downstream interpretation. Finally, we show an implementation of the model in NOCIt, a software system that calculates the a posteriori probability distribution on the number of contributors.

CONCLUSION

The model was selected using a large, diverse set of DNA samples obtained from 144 different laboratory conditions; with DNA amounts ranging from a single copy of DNA to hundreds of copies, and the quality of the profiles ranging from pristine to highly degraded. Implemented in NOCIt, the model enables a probabilisitc approach to estimating the number of contributors to complex, environmental samples.

STRmix™ collaborative exercise on DNA mixture interpretation

An intra and inter-laboratory study using the probabilistic genotyping (PG) software STRmix™ is reported. Two complex mixtures from the PROVEDIt set, analysed on an Applied Biosystems™ 3500 Series Genetic Analyzer, were selected. 174 participants responded. For Sample 1 (low template, in the order of 200 rfu for major contributors) five participants described the comparison as inconclusive with respect to the POI or excluded him. Where LRs were assigned, the point estimates ranging from 2 × 10⁴ to 8 × 10⁶. For Sample 2 (in the order of 2000 rfu for major contributors), LRs ranged from 2 × 10²⁸ to 2 × 10²⁹. Where LRs were calculated, the differences between participants can be attributed to (from largest to smallest impact): This study demonstrates a high level of repeatability and reproducibility among the participants. For those results that differed from the mode, the differences in LR were almost always minor or conservative.