Expansion of Microbial Forensics

Abstract

Microbial forensics has been defined as the discipline of applying scientific methods to the analysis of evidence related to bioterrorism, biocrimes, hoaxes, or the accidental release of a biological agent or toxin for attribution purposes. Over the past 15 years, technology, particularly massively parallel sequencing, and bioinformatics advances now allow the characterization of microorganisms for a variety of human forensic applications, such as human identification, body fluid characterization, postmortem interval estimation, and biocrimes involving tracking of infectious agents. Thus, microbial forensics should be more broadly described as the discipline of applying scientific methods to the analysis of microbial evidence in criminal and civil cases for investigative purposes.

INTRODUCTION

The anthrax letter attacks of 2001 in the United States ushered the world into a new reality and increased awareness of global vulnerability to bioterrorism. In addition, the event demonstrated that as a nation, the United States was woefully unprepared to characterize the biological evidence associated with the outbreak. As a consequence, the field of microbial forensics was developed to build a robust forensic capability to help investigate bioterrorism and biocrime. Microbial forensics is the discipline of applying scientific methods for the analysis of evidence from a bioterrorism attack, biocrime, hoax, or inadvertent release of a biological agent or toxin, with attribution as the ultimate goal (1). Attribution of microbial evidence involves determining an associated source and/or perpetrator or group of individuals to the highest degree possible. The microbial forensics field is built on a network of multiple specialties (e.g., microbiology, genetics, bioinformatics, forensic science, immunology, population genetics, biochemistry, molecular biology, epidemiology, etc.) and the law enforcement, public health, policy, and intelligence communities. While the field was formalized shortly after the anthrax letter attacks, its roots are well established, as they are embedded in the same practices used for decades in epidemiology and public health to investigate disease outbreaks. Epidemiologists focus on the outbreak, the population(s) at risk, spread of disease, possible reservoirs, and characterization of the etiologic agent (2), primarily serving the health care system. Epidemiology and microbial forensics are employed together to attempt to determine if an outbreak is natural, accidental, or intentional. Therefore, the two disciplines are integrated, and specialists tend to work together, with microbial forensics concentrating on individualization of the agent or toxin and/or how it was produced and disseminated. In addition, traditional forensic methods, such as fingerprinting, human DNA analysis, trace materials, and handwriting analysis, are exploited in a microbial forensic investigation, as ultimate attribution is the identification of the perpetrators of the crime.

There are more than 1,400 microbial species or strains that are potential health threats to humans (3), and the number expands by orders of magnitude when considering plant and animal pathogens. While high-consequence agents (4) have been targeted for developing preparedness measures, they are but a small percentage of possible biothreats, and it is not practical to continue to develop assays directed at single targets. There are simply far too many targets. Only 20 years ago, it took 13 months and cost >$870,000 for the first bacterial genome to be sequenced and assembled by the Institute of Genome Research (Rockville, MD) (5, 6). Seven years later for about the same amount of time and a lesser cost (approximately $200,000 to $300,000 for the first genome), genomic sequences of the Bacillus anthracis Ames strains were obtained from the evidence in the letter attacks and purported reference samples (7–9). Technical advancements in recent years, through the advent of massively parallel sequencing (MPS) (which also has been referred to as next-generation sequencing [NGS] and high-throughput sequencing [HTS]), allow analysis of microbes with a throughput and speed that were not thought possible a short time ago. MPS, a disruptive technology and a boon to microbial forensics, may overcome the challenge of identifying unknown pathogens, hoax microorganisms, and low-abundance microorganisms even in complex mixture samples. With its substantially increased throughput and continued development of powerful bioinformatics pipelines, MPS may be used to characterize any microbe, abundant or trace, degraded or intact, and even genetically engineered genomes with one unifying approach. MPS provides the ability to rapidly diagnose and monitor infections using culture-independent methods (thereby reducing cost and turnaround time) and track disease outbreaks in real-time using whole-genome comparisons (10–12). Indeed, Cummings et al. (13) showed several years ago the forensic capability of MPS to rapidly and reliably sequence multiple whole genomes. Since then, epidemiologists have applied MPS to several outbreak investigations (10–12, 14–17), and it is anticipated that MPS eventually will become the routine method for genetic analysis. In addition, MPS provides a methodology for human microbiome studies, which provide inference into different health and disease states and impact conditions, such as obesity, inflammatory bowel syndrome, effects from antibiotic use, and cancer (18–20). These same tools have been used to characterize the complex community of the human microbiome and have been demonstrated for use in human forensic applications, such as human identification, body fluid characterization, and time-since-death decomposition analysis.

Since the field of microbial forensics was developed in response to exigent circumstances, it was narrowly defined as concentrating on the immediate concern: bioterrorism. Other examples of microbial forensics investigations included tracing transmission of human immunodeficiency virus (HIV) and hepatitis C virus (HCV) in criminal health-related matters (21–28), which fall under the biocrime category (Table 1). Therefore, microbial forensics focused on investigations where the microbe or its products (e.g., toxins) were used as weapons or biothreats. However, over the past 15 years, technological advances and the realization of the vastness and abundance of the microbial world prompted the field to expand into other areas where microbes and their products may help other types of forensic investigations, including human identification (29) and postmortem interval estimation (30). This expansion in which microbes were exploited forensically beyond an investigation of bioterrorism and biocrime requires a more comprehensive definition for the field of microbial forensics. Microbial forensics now should be broadly defined as the discipline of characterizing microbiological evidence to develop investigative leads in criminal (and civil) cases.

TABLE 1.

Examples of microbial forensic and epidemiologic (of interest to forensics) investigations

Microbial type by yr	Case description	Reference(s)
Bacterial
2000–2001	Characterized Staphylococcus aureus strains from drug paraphernalia to track drug networks	66
Starting in 2002	Amerithrax investigation; compared different Ames strain morphotypes to determine the origin of the source material released in the 2001 anthrax attacks	107, 108
2009	Traced the origin of Bacillus anthracis from injectional anthrax cases among heroin users in Scotland	109
2010	Traced the origin of the source of the Haitian cholera outbreak after the 2010 earthquake	10, 110, 111
2011	Traced the origin of the 2011 Escherichia coli O104:H4 outbreak in Germany	11
Viral
1992	First report of molecular tracking of HIV infection from dentist to patients after invasive health care procedure	21
1994	Phylogenetic analysis to provide evidence that Richard Schmidt intentionally injected his girlfriend with HIV/HCV-contaminated blood	25
1998	Phylogenetic and molecular clock analysis to provide evidence that a Spanish anesthetist infected 275 patients with hepatitis C virus	28
2014	Traced the origin and transmission route of the 2014 Ebola outbreak	112
2015	Parvovirus B19 characterization from skeletal remains revealed the likely origin of World War II casualties	67

The expansion of the microbial forensics field extends into the realm of forensic human identity testing. Human microbiome analysis can be combined with traditional human DNA testing (e.g., short-tandem-repeat [STR] and single nucleotide polymorphism [SNP] analyses) to potentially provide additional data for stronger associations and better chances to exclude individuals falsely associated with biological evidence (Fig. 1). Human DNA-based identity testing allows for an analysis of stable inherited markers in the human genome related to individualization, kinship, ancestry, and phenotype. In contrast, forensic microbiome testing allows for the analysis of both stable and fluctuating changes in microbial communities on and in the body related to individualization, diet, health, recent geolocation, and postmortem time intervals. Microbial genetic markers can be used to expand upon current forensic genetic testing capabilities.

FIG 1.

Diagram of expanded human investigative and forensic testing to include human genome and human microbiome characterization.

MICROBIAL FORENSICS AND HUMAN IDENTITY TESTING

Humans are born with approximately 20,500 genes and die with more than 1,000,000 genes (31–33). This change in gene count is due to the accumulation of microorganisms as part of the normal development and existence of human beings. Indeed, there are 10 times more bacterial cells in and on the body than there are human cells, making us reconsider what makes us human. The substantial microflora that humans carry is known collectively as the human microbiome (32, 33). The majority of human microbes are autonomous, self-replicating, transmissible, unavoidable, and, in general, ubiquitous, although they may vary to some degree or substantially from human to human (34, 35). The microorganisms and their accompanying nucleic acids that are carried by humans are shed, deposited, and exchanged routinely in a similar fashion to human DNA, which is exploited for identifying people involved in sexual assaults, murders, and burglaries. These microbial genomes from many different microbial species collectively are complex and variable and may provide forensic signatures that could be as individualizing as are the current set of human short-tandem-repeat markers (36) used for identity testing purposes. In addition, since microorganisms that reside in different areas of the human body vary, additional investigative value could be obtained for determining the tissue source of forensic biological evidence. Thus, the human microbiome may be another target that could be used to identify (or exclude) humans involved in crimes.

When people touch items, they often transfer their DNA onto objects (via primary, secondary, and tertiary transfer) (37, 38). Therefore, an individualizing signature is left behind that can be exploited to determine the identity of an individual who may have handled an object. Forensic DNA typing characterizes genetic signatures from human biological samples. The current DNA typing methodologies focus on markers in the human genome and are sensitive, highly discriminating, and well validated (for examples, see references 39–41). Unfortunately, the amount of human DNA deposited by touching an object often is very low, and most technologies cannot reliably type such low levels of DNA. To attempt to obtain results from such limited samples, modifications of current methods are made to increase the sensitivity of detection of human DNA. Collectively, the suite of methods that increase the sensitivity of human DNA typing protocols are known as low-copy-number (LCN) typing (42–44). Under LCN typing, limited template analyses suffer from exaggerated stochastic effects, and with increased sensitivity, there is a greater potential for contamination. This lack of reproducibility with LCN typing results has not deterred the interest in LCN typing and its potential use for developing investigative lead purposes. One way to partially overcome the uncertainty of the LCN typing methodology is to employ an orthogonal approach. In fact, each technology, i.e., LCN typing and the orthogonal methodology, may not be robust, but when performed together, the information may be sufficiently corroborating to decrease uncertainty, enabling better use results from trace biological sample analyses for developing investigative leads.

Given the much greater number of bacterial cells compared with human cells, it is conceivable that more bacterial cells and thus gene targets are deposited on touched items than are human markers. Indeed, Grice et al. (45) showed that 10,000 bacteria/cm² and 50,000 bacteria/cm² could be collected by swabbing and scraping the skin, respectively. Of course, the population of bacterial cells is composed of many species with various abundance levels. However, a combination of enrichment methods, such as PCR, and sequencing can be used to detect those species that may be used to individualize their human hosts. Goga (46) sought to analyze bacterial DNA from shoes, as the quantity of human DNA there often was at LCN typing levels. Goga demonstrated that there are sufficient bacteria in shoes and the plantar skins of individuals and that the microbiome communities were unique among the individuals tested. A level of “matching” was possible between the shoe and wearer profiles. However, the results varied substantially, and in a few comparisons, the identity of the wearer and another individual could not be resolved. Tims et al. (47) analyzed the microbiome before and after hand washing and observed similar variation. They opined that constant bacterial contamination by touching may impact the results. These findings were not similar to those of Fierer et al. (29, 48), who collected reference bacterial samples without washing and successfully matched reference microbial communities to those deposited on touched objects, such as keyboards and computer mice. Fierer et al. (48) suggested that washing may have changed the composition of the hand skin microbiome. For hand microbiome profiling, an unwashed hand may be a better starting point for studies. More studies are needed to determine the suitability of touch sample analyses.

Gut microbiome profiling also may be an important tool for forensic human identity testing. There is evidence that core gut microflora of an individual are stable (49, 50) but can be affected by environmental changes and antibiotic use (51, 52). Fecal material has been found at some crime scenes, and determining the source of such evidence may be an important investigative lead. Typically, human STR profiling is attempted but with mixed success. However, fecal material, the primary material used to study the gut microbiome, carries a microbiome profile that could be exploited for identification purposes. The stable microbiome may provide identity information, while fluctuating or transient microbiota might indicate recent diet or geolocation.

Identification of the tissue source of forensic biological samples can be critical in some investigations to reconstruct crime scenes and events, but current techniques are limited. Presumptive tests for the specific body fluids are used as screening tests and tend to have specificity limitations. Screening, which is often quick and inexpensive, is used to select the best candidate samples for more-in-depth testing. Confirmatory tests identify tissues with high specificity. While gene expression (mRNA typing) (53, 54) and methylation (55) have been described, most presumptive and confirmatory tests are protein-based enzymatic or immunologic assays. The exception is microscopic visualization of sperm for the identification of semen. Proteins are less stable than DNA, and a negative result may not indicate reliably the tissue source of a sample or the amount of DNA in a sample. Most tissue assays are non- or semiquantitative without a defined threshold for a positive reaction. Because of the different assay formats for each tissue-specific target, presumptive and confirmatory tests cannot be run in a similar parallel manner, adding to labor demand and the difficulty of automation in a cost-effective way.

Unique bacterial genetic signatures may aid in biological sample tissue source determination. For example, Lactobacillus crispatus, Lactobacillus jensenii, and Atopobium vaginae have been associated with vaginal secretions, while Lactobacillus iners, Lactobacillus gasseri, and Gardnerella vaginalis have been found in other body fluids as well (56–58). Tissue specific to the vagina could be informative in rape cases. Nakanishi et al. (59) showed that Streptococcus salivarius and Streptococcus mutans could be detected in mock forensic saliva samples and were not present in other forensically relevant tissue sources. Choi et al. (60) had similar findings for S. salivarius. These species are relatively abundant and thus may be easily detected.

Nose and throat (respiratory and digestive) commensal bacteria are likely candidates for geographical and recent contact information because they demonstrate high temporal stability (persistence), are contagious organisms (airborne and fomites), exhibit clonal and geographical variation and diversity, colonize a large portion of the population, and are easy to collect (e.g., see references 61 and 62). Scheidegger and Zimmerli (63) have shown that Staphylococcus aureus is a high-incidence bacterium among drug users. Shared drug paraphernalia can be an avenue for bacterial transmission (64), in which colonization can be transient or permanent (65). A study of inhalational drug users detected similar S. aureus strains between users and drug paraphernalia (66). The S. aureus strain analysis traced back 14 social networks (known groups of individuals connected by social interaction and reported drug use) linked to a crack house, while only two of the biological networks (nasal culture and drug paraphernalia linked by significant similarity among isolates) were identified by social network analysis. These data demonstrate that behavioral, social, and environmental networks may be identified by analysis of a targeted species of the human microbiome. Because environment is related to social networks, hygiene, diet, geography, etc., microbiome analyses could be an additional tool for intelligence gathering. Last, the profiles could be used to confirm or refute what persons of interest claim about where they have been and with whom they have associated. If the microbial profile of an individual reflects the unique nature of a complex microbial community, microbial profiles may be better than human DNA typing for recent geolocation, as the environment (including contact with groups of individuals) may carry location-specific microorganisms that will be exchanged through human contact.

Tracking sexually transmitted diseases in child molestation, rape cases, and questioned malpractice cases in health care provider outbreaks have been well documented for HIV and HCV cases (21–28) and have been part of the microbial forensics arena. A full-scale analysis that demonstrated the potential of such investigations was reported by Gonzáles-Candelas et al. (28). These authors described in detail a case of a Spanish anesthetist convicted of malpractice by infecting more than 270 of his patients with hepatitis C virus (HCV). The case is particularly interesting due the time frame of the investigation some 25 years after the first suspected transmission(s) and by the number and complexity of potential infectious events (>300 candidates from two hospitals). Thus, scientifically challenging but forensically obvious questions were posed to the experts. (i) Was the suspect the source responsible for the outbreak? (ii) Could it be ascertained whether the patients that had been infected share a source and thus could be included in the outbreak? (iii) Alternatively, which patients could have been infected from other sources? (iv) Could these alternative sources or the existence of different but simultaneous outbreaks be determined? (v) Could the duration of the outbreak be determined? (vi) Could the time of infection for each patient in the outbreak be estimated? Finally, (vii) could the date of infection of the anesthetist be estimated? These questions were tackled by sequencing 229 nucleotides of a conservative nonstructural 5B (NS5B) gene using standard Sanger sequencing technology. After systematic and rigorous analyses, the authors concluded that the suspected health care professional was indeed the source of the HCV infection of a portion of infected individuals and provided likelihood ratios to support the findings versus an alternative hypothesis. As in many other forensic DNA investigations, the data were not used to prove if the suspect was guilty, as the results provide only the identity of the virus and strain (or quasispecies). The data combined with other evidence were used to convict the anesthetist.

The human virome also potentially can be used for characterizing unidentified cadavers, even in cases of skeletonized remains. Toppinen et al. (67) are the first to describe the suitability of bone as a source for exploration of DNA viruses (as ancient DNA researchers have described bacterial strains of plague victims from past pandemics [68–70]) but instead for forensic identity purposes. They showed parvovirus B19 DNA sequences were found abundantly in 70-year-old long bones from putative casualties from World War II. The reported viral sequences were exclusively from one genotype, which disappeared from circulation in the 1970s, or from a genotype that has never been reported in northern Europe. By adding the viral information to that from human mitochondrial and Y-chromosome profiling, the authors concluded a dispute of the origin of two individuals found in the battlefield between Finland and Russia.

These cases provide insight into the potential of exploiting the microbial world for forensic purposes. It is clear that the human microbiome holds substantial information to assist the forensic science community's analysis of a variety of case scenarios.

MPS APPROACHES

Most microbial analysis methods, and in particular metagenomics, have focused on the single target of the 16S rRNA gene, which lacks species-level resolution. Some of the above-mentioned studies targeted only a few known species that may be unique to a body fluid or tissue, such as saliva or vaginal secretions. They only require a positive/negative result for tissue sourcing. However, most of the studies were not sufficient to unequivocally identify the targeted species among a complex metagenomics background necessary for forensic investigations, and generally, they were not sufficient to provide high-depth characterization at the subspecies or genotype level. Lastly, such tissue-sourcing assays often require the redesign of primers to increase specificity and continual evaluation of potential false positives with increased sampling and databases as they are updated with new genomic sequences. If specialized primer design is required for the detection of every, or most, microorganism of interest, there will be great demand on technology development and require substantial and costly validation.

Current major genetic and health efforts describe the diversity of the human microbiome (18, 32, 35, 45, 71–73). However, little work has been carried out on how individualizing the profiles may be, which microorganisms are common to all humans, what portion of these common human microbiome species are stable throughout an individual's lifetime (or at least for reasonably long time periods), and what portions vary due to environmental conditions. Such questions cannot be answered until (i) a defined set of resolving markers are developed to assay the human microbiome, and (ii) a defined set of bacterial species are determined that are best suited for human identity testing.

There are technical challenges to identify bacteria of interest within highly complex metagenomic samples, to distinguish those of interest from near neighbors and from the vast complex background that constitutes a microbial sample taken from body areas, and the degree of confidence that can be assigned to a potential species that is detected in such complex samples. One approach to meet these objectives is to develop a panel of bacterial markers that have the potential to differentiate human hosts and that can be multiplexed in a MPS system. A targeted bacterial marker panel would provide an efficient and cost-effective method that balances sequence coverage and throughput and yet provides species-level resolution. MPS could provide a single unifying methodology so the power of the assay can be realized and resources for testing and validation can be used in a cost-efficient manner. Sensitive and accurate bacterial DNA detection is imperative for forensics analyses. Validation is an essential part of any assay and should be become a routine part of technology development and implementation (see below). The aim to achieve the goals of human identification through the microbiome is essentially targeted genomics to survey the variety of microorganisms present in a specific sample. Microbes of interest for human identity testing should be common to all healthy individuals and sufficiently abundant to be detected routinely. Human microbiomes represent unique ecosystems composed of complex mixtures of microorganisms. For the needs of forensic diagnostics, analysis of metagenomic samples can be more focused than total community structure and organism abundance studies. Instead, molecular analyses can entail the identification of specific microorganisms that provide an individualizing signature to the highest degree possible. Such information requires the resolution of selected microorganisms from near neighbors as well as from all other microorganisms of the community that create noise and reduce the sensitivity of detection of an assay.

Most metagenomic analyses characterize whole microbial communities at the phylum level, often reporting different abundance ratios of partially resolved taxa. While interesting and informative for ecosystem analyses, that level of resolution and the variation in abundance from sample-to-sample likely will not achieve human individualization. Instead, for human identification, it is imperative to (i) identify key bacteria that are common to all individuals so identity testing can be informative with a routine target set of microorganisms, (ii) select those bacteria that are relatively abundant to reduce stochastic sampling effects, and (iii) target genes or sequences that contain sufficient variation to generate a profile that would provide a high degree of individualization of the donor(s) of bacterium-containing biological evidence.

Current metagenomic approaches apply MPS and target a single phylogenetic marker, the 16S rRNA gene, or perform whole-genome shotgun sequencing. MPS with the 16S rRNA gene provides deeper coverage but rarely can differentiate at the species level. Whole-genome shotgun sequencing may be able to differentiate at the species level but lacks depth of coverage, and thus, stochastic effects reduce its ability to achieve taxonomic resolution. The 16S rRNA gene is the most commonly used bacterial genetic marker in phylogenetic studies and broad bacterial identification. The conserved and variable regions of the gene, its presence in a number of databases (e.g., see references 74–76), and the significant volume of 16S rRNA studies add to the appeal of using this marker. However, there are limitations to solely using 16S rRNA, which include insufficient species resolution (77), PCR bias (78, 79), copy number variation (80), and sequence variability among a single bacterium (81), inaccurate phylogenetic relationships based on key variability outside the marker region (82), and horizontal transfer of the entire gene region (83, 84). These limitations of using a single gene target to identify bacteria in a complex community will lead to inaccurate abundance ratios and can confound phylogenetic analyses. Whole-genome shotgun sequencing provides the ability theoretically to sequence all DNA molecules in a given sample, potentially covering any given region(s) of many genomes. Whole-genome shotgun sequencing can obtain species- or strain-level characterization of a given genome by producing sequence reads of species-/strain-specific informative markers. However, the more area of any given genome that is attempted to be sequenced, the less depth of coverage that will be obtained for any particular site, potentially reducing the confidence for species identification. Highly complex metagenomic samples can contain thousands of species within a sample, thereby limiting the coverage of any one genome, especially those at low abundance, such as may be inherent to trace biological samples.

A novel metagenomics approach that employs the use of multiple informative phylogenetic markers can allow greater depth of coverage of targeted genomic regions than would be achieved by whole-genome shotgun sequencing. The markers suited for the task of species-level resolution are housekeeping genes, as they contain appropriate conserved and variable regions to differentiate among species. The use of housekeeping genes for bacterial species- and strain-level identification is not a new concept. Multilocus sequence typing (MLST), first described almost 20 years ago (85), is one of the most commonly used bacterial typing methods. MLST typically uses seven housekeeping genes, and the composite locus sequence creates a profile for comparison purposes. Online MLST databases are available for identification purposes and data storage (86–88). With MPS, an augmented MLST method has been developed using 20 genes (or 21 genes [89]) called MLST-seq (90). In addition, bioinformatics software packages are freely available to analyze MLST profiles from shotgun sequencing data (88, 91, 92). Even with these advances, there are limitations with the MLST method, which include nonuniversal gene panels and insufficient species-level resolution. Seven markers simply are not sufficient for identification to the species level. In these cases, multilocus variable-number tandem-repeat analysis (MLVA) and single nucleotide polymorphism (SNP) analyses have proved beneficial for additional resolution of specific species (93). An expanded MLST approach has been developed, called ribosomal multilocus sequence typing (rMLST), utilizing 53 housekeeping genes to type bacteria down to the subspecies level (94). Another method using housekeeping genes as genetic markers was described by Baldwin et al. (95). Their approach consisted of a PCR-based assay using 16 different primer pairs for amplifying 9 housekeeping genes, followed by electrospray ionization mass spectrometry for genus- and species-level characterization (95). This mass spectrometry-based assay is very appealing and supports the concept of multiple markers for identification to the species level, but it is limited to detecting only the most abundant 1 to 3 species in a sample, i.e., it is better suited for testing infected individuals with a high titer of the target species.

Bioinformatics programs can retrieve and use housekeeping genes as genetic markers from shotgun sequencing data. AMPHORA (AutoMated PHylogenOmic infeRence) has been developed using a panel of 31 housekeeping genes for better taxonomic resolution than solely using 16S rRNA (96). These genes were selected because they are found within all bacteria, mostly present in single copies, and believed to be fairly resistant to horizontal gene transfer (96). This 31-housekeeping-gene panel was modified, and another bioinformatic program, called MetaPhyler, was developed to analyze the data (97). Phyla-AMPHORA was developed utilizing thousands of phylum-specific phylogenetic markers to improve the resolution of phylogenomic analyses over that with the initial 31 genes (98). Additionally, PhyloSift enables phylogenetic analysis of metagenomes, which expands on AMPHORA, and includes 37 bacterial and archaeal genetic markers (PhyEco markers) (99) in the prokaryote core marker set (100). PhyloSift also provides capabilities for the use of extended and custom marker sets and has software packages for metagenomic data set simulation (to test newly generated custom markers) and statistical analyses (100). Another bioinformatics tool, mOTU, was developed using 40 marker genes (101) to profile metagenomes with species-level resolution (102). Additionally, FunGene (the Functional Gene Pipeline and Repository) includes 11 phylogenetic markers in its repository and analysis pipeline (103). Although these programs use different genetic markers (with a subset of overlapping markers), they demonstrate that the technology and software exist to develop a custom genetic panel consisting of informative phylogenetic markers to use for species-level identification in human microbiome samples.

VALIDATION

Validation is essential in the development of diagnostic methods, and those used in microbial forensics are no exception. The most important consideration is that microbial forensic methods may develop information that can impact the health, life, and freedom of individuals, policy decisions, and possibly action by governments on a large scale. Therefore, the results generated from the analyses of any microbial forensics analysis (including collection, transfer, analytical method, and interpretation, as well as proper training) must be understood and limitations defined, so that a proper degree of confidence can be associated with the findings. Accurate and credible results are a requisite. Budowle et al. (104) stressed that failure to properly validate a method or misinterpret the results from a microbial forensic analysis or process may have severe consequences.

Validation is the process that “1) assesses the ability of procedures to obtain reliable results under defined conditions, 2) rigorously defines the conditions that are required to obtain the results, 3) determines the limitations of the procedures, 4) identifies aspects of the analysis that must be monitored and controlled, and 5) forms the basis for development of interpretation guidelines to convey the significance of the findings” (104). The basic features and criteria for validation are listed in quality assurance guidelines for microbial forensics (1) and are addressed elsewhere in detail (104, 105). Budowle et al. (104) define developmental validation as “the acquisition of test data and the determination of conditions and limitations of a newly developed method to analyze samples” and internal validation “as an accumulation of test data within the operational laboratory to demonstrate that established methods and procedures perform within predetermined limits in the laboratory.” Every effort should be made to validate a method thoroughly before implementation. However, cost, resources, and exigent circumstances may require a method to be implemented without extensive validation. Clearly, it is unacceptable to hold off on any analyses because an available method has not been completely validated, especially when an attack is under way. In this context, Schutzer et al. (106) described the process of preliminary validation as “an early evaluation of a method that will be used to investigate a biocrime or bioterrorism event.” There is still a requirement to acquire some test data to evaluate a potential method so peer review of extant data can be performed quickly by a panel of experts. This panel would recommend limitations and additional studies (if necessary) to use prior to analyzing evidentiary material. However, it is stressed that a preliminary validation should not be used as an excuse for obviating a full validation.

Budowle et al. (105) recently described criteria for the validation of MPS procedures. The criteria are not novel, as the same approaches are being used for human genome sequencing and clinical diagnostics. Three common areas of MPS validation are (i) sample preparation, (ii) sequencing, and (iii) data analysis, but these criteria should be defined in terms of the specific application. There are a number of general topics subsumed within these three areas, which are listed in Table 2.

TABLE 2.

Validation criteria for MPS procedures^a

Validation criteria
Extraction of DNA (or RNA)
Quality
Purity
Enrichment
PCR
Capture
Library preparation
Multiplexing
Markers
Samples
Sequencing
Data analysis
Raw data processing
Quality scores
Alignment
Variant calls
Reference materials
Test materials
Controls
Databases
Test materials
Inferences of results
Bioinformatics
Software tools
Data management
Data storage
Interpretation
Taxonomic assignment
Abundance
Organism classification
Community structure
Standard operating protocols
Reporting

^aBased on data from reference 105.

Addressing these general areas of the MPS method should provide an analytical tool that is sufficiently robust for the expanding field of microbial forensics. While the field certainly has matured and no longer can be considered solely the domain of bioterrorism and biocrime investigations, MPS technologies are still quite dynamic. It is anticipated that the speed of obtaining sequencing results will increase, and read lengths will be extended, with a concomitant decrease in cost. Applications beyond those discussed here will become part of the microbial forensics arena. Good-quality practices will ensure that the tools of microbial forensics and the results obtained will be robust, accurate, and reliable.

As the microbial forensics field expands to incorporate the use of microbiome characterization for human investigative and forensic testing, additional methods can be created to expand the forensic genetic toolbox. Likely, more-advanced sequencing methods can be developed to better elucidate the data that can be obtained from complex metagenomic samples, which can provide additional investigative value, including geolocation, infection source tracking of biocrimes, postmortem interval, and increased power of discrimination for human identification. As standards and quality practices are developed for MPS and metagenomic microbial forensic testing methods, human microbiome characterization likely will become a routine methodology in the field of microbial forensics.

Biographies

Sarah E. Schmedes, M.S., is a Ph.D. Candidate in the Department of Molecular and Medical Genetics at the University of North Texas Health Science Center (UNTHSC) in Fort Worth, TX. She earned her Bachelor of Science in biochemistry and microbiology from Texas State University in San Marcos, TX, and earned her Master of Science in forensic biology from the University of Albany, State University of New York. Prior to enrolling in the molecular genetics doctoral program at UNTHSC, she served as a forensic research assistant at the Institute of Applied Genetics at UNTHSC. She has published 8 manuscripts and 1 book chapter during her graduate career. Sarah is a student member of the American Society for Microbiology and was a recipient of the 2014 Eugene and Millicent Goldschmidt Graduate Student Award, awarded by the Texas Branch-American Society for Microbiology. Her research interests include metagenomics, microbial forensics, and emerging infectious diseases.

Antti Sajantila, M.D, Ph.D., is a trained forensic pathologist and forensic geneticist from the Department of Forensic Medicine, University of Helsinki, Finland. He received his Ph.D. from the University of Helsinki, Finland, and was a postdoctoral fellow in Ludwig Maximilian's University in Munich, Germany. He has published over 200 articles, book chapters, and reviews. His research interests are in the fields of human population genetics, human identification, disaster victim identification, and cause-of-death investigation. He has testified in forensic court cases in Finland, Australia, and Kenya. He also has served on several national and international committees for the application of DNA methods in human identification. He is currently a member of Interpol's Disaster Victim Identification Standing Committee's Forensic Pathology and Forensic Genetics Working Groups.

Bruce Budowle received a Ph.D. in genetics in 1979 from VA Tech. In 1983, he joined the FBI Laboratory Division to carry out research, development, and validation of methods for forensic biological analyses. He has published approximately 550 articles and testified in well over 250 criminal cases. He has authored or coauthored books on molecular biology techniques, electrophoresis, protein detection, and microbial forensics. He has been directly involved in developing quality assurance standards for the forensic DNA field. Some of his efforts over the last 20 years have been in microbial forensics and bioterrorism. In 2009, B. Budowle became Executive Director of the Institute of Applied Genetics and Professor in the Department of Molecular and Medical Genetics at the University of North Texas Health Science Center at Fort Worth, TX. His current efforts focus on human forensic identification, microbial forensics, and emerging infectious disease.