A probabilistic approach to evaluate salivary microbiome in forensic science when the Defense says: `It is my twin brother'

Highlights

• salivary microbiota composition to support discrimination between related individuals.

• Beta-diversity indeces to highlight taxonomical differences between pairs of individuals.

• a probabilistic approach to evaluate salivary microbiome.

Abstract

Salivary microbiota profiles may represent a valid contribution to forensic investigation when standard DNA genotyping methods fail. Starting from questioned and control materials in the form of saliva, the evidence can be expressed by means of a distance between those materials taking into account specific aspects of the microbiota composition. The value of the evidence for forensic discrimination purposes is quantified by means of a Bayes’ factor, that allows one to overcome the major limitations and pitfalls of intuition connected to the use of cut-off values as a mean of decision.

Introduction

The suitability of salivary microbiota composition to support discrimination between closely related individuals has been investigated in [17]. Microbiota profiles have shown high individuality and stability over time and may represent a valid and precious contribution for forensic investigations when standard genotyping DNA methods fail (e.g. due to degraded material or to close relatedness among individuals, such as homozygous twins). Previous studies have used microbiomes to differentiate an individual from a limited group of people (see, e.g. [6]).

Beta-diversity (β-diversity) indices (such as the Jaccard distance) are commonly used in microbiota studies to highlight taxonomical differences between pairs of samples [10]. They have been implemented to overcome the dimensionality problem that renders the quantification of the ressemblance between microbiota profiles problematic. Their adequacy and feasibility for discrimination between hypotheses of forensic interest have been discussed in [17].

The assessment of the value of a score quantifying the similarity between measurements originating from questioned material and from control material for discrimination purposes is an open problem, as reiterated by [6]. The use of a cut-off value against which such score is compared in order to support or even suggest a conclusion regarding their origin (e.g. whether compared saliva traces originate from the same individual or from closely related individuals) is often suggested. Routine examples come mainly from the toxicology domain where a scientist is often asked to classify a controlled substance in a blood or hair item into a predefined category (e.g. doping athletes, abusive drinkers, and so on). However, as it will be discussed in Section 5, such an approach presents severe limitations and leaves the problem of forensic interpretation open. Most importantly, it renders a binary response, no matter the effective similarity (as measured by a beta-diversity indece) between the compared material. In the current paper, a probabilistic approach to evidence evaluation is proposed. The most efficient way to provide a quantitative measure of the support of the evidence (i.e., a β-diversity distance) to competing hypotheses about the origin of a given salivary material is given by the Bayes’ factor that expresses, in probabilistic form, the ratio between the probability to observe the evidence given each of the hypotheses of interest.

The paper is structured as follows. The scenario of interest and the proposed probabilistic approach are introduced in Section 2. Available data and the chosen statistical models are described in Section 3, while the results of statistical analyses are presented and discussed in Section 4. Section 5, finally, concludes the paper.