Yuya Hamano

Forensic age prediction for dead or living samples by use of methylation-sensitive high resolution melting

Age prediction with epigenetic information is now edging closer to practical use in forensic community. Many age-related CpG (AR-CpG) sites have proven useful in predicting age in pyrosequencing or DNA chip analyses. In this study, a wide range methylation status in the ELOVL2 and FHL2 promoter regions were detected with methylation-sensitive high resolution melting (MS-HRM) in a labor-, time-, and cost-effective manner. Non-linear-distributions of methylation status and chronological age were newly fitted to the logistic curve. Notably, these distributions were revealed to be similar in 22 living blood samples and 52 dead blood samples. Therefore, the difference of methylation status between living and dead samples suggested to be ignorable by MS-HRM. Additionally, the information from ELOVL2 and FHL2 were integrated into a logistic curve fitting model to develop a final predictive model through the multivariate linear regression of logit-linked methylation rates and chronological age with adjusted R(2)=0.83. Mean absolute deviation (MAD) was 7.44 for 74 training set and 7.71 for 30 additional independent test set, indicating that the final predicting model is accurate. This suggests that our MS-HRM-based method has great potential in predicting actual forensic age.

Forensic age prediction for saliva samples using methylation-sensitive high resolution melting: exploratory application for cigarette butts

There is high demand for forensic age prediction in actual crime investigations. In this study, a novel age prediction model for saliva samples using methylation-sensitive high resolution melting (MS-HRM) was developed. The methylation profiles of ELOVL2 and EDARADD showed high correlations with age and were used to predict age with support vector regression. ELOVL2 was first reported as an age predictive marker for saliva samples. The prediction model showed high accuracy with a mean absolute deviation (MAD) from chronological age of 5.96 years among 197 training samples. The model was further validated with an additional 50 test samples (MAD = 6.25). In addition, the age prediction model was applied to saliva extracted from seven cigarette butts, as in an actual crime scene. The MAD (7.65 years) for these samples was slightly higher than that of intact saliva samples. A smoking habit or the ingredients of cigarettes themselves did not significantly affect the prediction model and could be ignored. MS-HRM provides a quick (2 hours) and cost-effective (95% decreased compared to that of DNA chips) method of analysis. Thus, this study may provide a novel strategy for predicting the age of a person of interest in actual crime scene investigations.

Development and validation of open-source software for DNA mixture interpretation based on a quantitative continuous model

In criminal investigations, forensic scientists need to evaluate DNA mixtures. The estimation of the number of contributors and evaluation of the contribution of a person of interest (POI) from these samples are challenging. In this study, we developed a new open-source software “Kongoh” for interpreting DNA mixture based on a quantitative continuous model. The model uses quantitative information of peak heights in the DNA profile and considers the effect of artifacts and allelic drop-out. By using this software, the likelihoods of 1–4 persons’ contributions are calculated, and the most optimal number of contributors is automatically determined; this differs from other open-source software. Therefore, we can eliminate the need to manually determine the number of contributors before the analysis. Kongoh also considers allele- or locus-specific effects of biological parameters based on the experimental data. We then validated Kongoh by calculating the likelihood ratio (LR) of a POI’s contribution in true contributors and non-contributors by using 2–4 person mixtures analyzed through a 15 short tandem repeat typing system. Most LR values obtained from Kongoh during true-contributor testing strongly supported the POI’s contribution even for small amounts or degraded DNA samples. Kongoh correctly rejected a false hypothesis in the non-contributor testing, generated reproducible LR values, and demonstrated higher accuracy of the estimated number of contributors than another software based on the quantitative continuous model. Therefore, Kongoh is useful in accurately interpreting DNA evidence like mixtures and small amounts or degraded DNA samples.

Pairwise Kinship Analysis by the Index of Chromosome Sharing Using High-Density Single Nucleotide Polymorphisms

We developed a new approach for pairwise kinship analysis in forensic genetics based on chromosomal sharing between two individuals. Here, we defined “index of chromosome sharing” (ICS) calculated using 174,254 single nucleotide polymorphism (SNP) loci typed by SNP microarray and genetic length of the shared segments from the genotypes of two individuals. To investigate the expected ICS distributions from first- to fifth-degree relatives and unrelated pairs, we used computationally generated genotypes to consider the effect of linkage disequilibrium and recombination. The distributions were used for probabilistic evaluation of the pairwise kinship analysis, such as likelihood ratio (LR) or posterior probability, without allele frequencies and haplotype frequencies. Using our method, all actual sample pairs from volunteers showed significantly high LR values (i.e., ≥ 108); therefore, we can distinguish distant relationships (up to the fifth-degree) from unrelated pairs based on LR. Moreover, we can determine accurate degrees of kinship in up to third-degree relationships with a probability of > 80% using the criterion of posterior probability ≥ 0.90, even if the kinship of the pair is totally unpredictable. This approach greatly improves pairwise kinship analysis of distant relationships, specifically in cases involving identification of disaster victims or missing persons.

New stutter ratio distribution for DNA mixture interpretation based on a continuous model.

In forensic science, DNA mixture interpretation is traditionally based on a binary model, which does not account for peak-height information in DNA profiles. In recent years, some countries have adopted a continuous model in which peak heights are used and stochastic effects are considered to enable rigorous calculation of likelihood ratios. However, this model requires certain biological parameters which affect the expected allelic and stutter peak heights. In this paper, we focused on estimating the distribution of the stutter ratio (SR) in 15 short tandem repeat loci in relation to the allele repeat number. We estimated the SR values of 234 single-source DNA samples by using a commercially available kit. In all loci except for D8S1179, D21S11, and D2S1338, a simple log-normal distribution model was fitted to the variability of SR. For D21S11, we developed a new distribution model in which distinct log-normal distributions between complete and incomplete repeat units are used (a separate log-normal distribution model). For D8S1179 and D2S1338, we developed another new distribution model that mixes two log-normal distributions to explain two types of repeat structures appearing within the same number of allele repeats. These two models were fitted to the observed SR values more accurately than the simple log-normal distribution model. We expected these new SR models to be applied to DNA mixture interpretation based on a continuous model.

Optimal small-molecular reference RNA for RT-qPCR-based body fluid identification

MicroRNA (miRNA) -based body fluid identification (BFID) plays a prominent role in a forensic practice, and the selected reference RNA is indispensable for a robust normalization in BFID performed using reverse transcription-quantitative PCR. In this study, we first examined sample quality using RNA integrity number, then evaluated the consistency of expression of candidate reference RNAs in 4 forensically relevant body fluids using NormFinder and BestKeeper, and lastly used each rank and index output from these tools for selecting the optimal reference RNA and the combination of the multiple RNAs using the RankAggreg package of R. We found that RNA integrity number was small in our samples, despite the use of pristine body fluids; 5S-rRNA was the optimal reference RNA for the identification of forensically relevant body fluids; and the combination of 5S-rRNA and miR-92a-3p and/or miR-484 enhanced the normalization quality. Our findings enable us to perform stringent normalization of the expression of body fluid-specific RNAs, and thus, can contribute to the development of small RNA-based BFID systems.

Discrimination of relationships with the same degree of kinship using chromosomal sharing patterns estimated from high-density SNPs

Distinguishing relationships with the same degree of kinship (e.g., uncle-nephew and grandfather-grandson) is generally difficult in forensic genetics by using the commonly employed short tandem repeat loci. In this study, we developed a new method for discerning such relationships between two individuals by examining the number of chromosomal shared segments estimated from high-density single nucleotide polymorphisms (SNPs). We computationally generated second-degree kinships (i.e., uncle-nephew and grandfather-grandson) and third-degree kinships (i.e., first cousins and great-grandfather-great-grandson) for 174,254 autosomal SNPs considering the effect of linkage disequilibrium and recombination for each SNP. We investigated shared chromosomal segments between two individuals that were estimated based on identity by state regions. We then counted the number of segments in each pair. Based on our results, the number of shared chromosomal segments in collateral relationships was larger than that in lineal relationships with both the second-degree and third-degree kinships. This was probably caused by differences involving chromosomal transitions and recombination between relationships. As we probabilistically evaluated the relationships between simulated pairs based on the number of shared segments using logistic regression, we could determine accurate relationships in >90% of second-degree relatives and >70% of third-degree relatives, using a probability criterion for the relationship ≥0.9. Furthermore, we could judge the true relationships of actual sample pairs from volunteers, as well as simulated data. Therefore, this method can be useful for discerning relationships between two individuals with the same degree of kinship.