NEW YORK (GenomeWeb) – Metagenomic sequencing methods are increasingly being explored for soil forensics as investigators push for ways to use microbial DNA profiles to differentiate similar soils from distinct locales.
In a paper published earlier this month in PLOS One, researchers from Flinders University and Forensic Science South Australia compared and contrasted three strategies for applying random metagenomic sequencing methods forensically to source soil samples from two residential parks found a couple miles apart.
The first involved direct shotgun sequencing on the complete collection of microbial DNA in each soil sample, senior author Adrian Linacre, a biological sciences researcher at Flinders University, told In Sequence in an email message.
Another approach included a whole-genome amplification step prior to metagenomic sequencing of DNA from soil microbes, Linacre explained, while a third method relied on PCR amplification with an arbitrarily selected primer that targeted DNA with GC-rich sequences.
While the techniques provided clues to the microbes, microbial genes, and metabolic pathways present in a given sample, the team found that the arbitrarily primed PCR, or AP-PCR, approach offered the most accurate and reproducible discrimination between samples from different sites.
"Our data showed that shotgun and WGA-based approaches generated highly similar metagenomic profiles … such that the soil samples could not be distinguished accurately," Linacre and his co-authors noted, while the strategy involving AP-PCR amplification was "successful at obtaining reproducible site-specific metagenomic DNA profiles, which in turn were employed for successful discrimination of visually similar soils samples collected from two different locations."
Based on their results so far, the researchers plan to apply sequencing to soil samples from both mock and real-life cases to determine whether the performance achieved in the current study holds up in other situations.
The potential utility of soil material in a wide range of forensics settings has prompted interest in finding accurate and reliable strategies for characterizing soil and matching it to its source location, the researchers noted.
"Soil, owing to its inherent features, adheres under fingernails, to cars, tools, weapons, or items of clothing and can transfer during the commission of a criminal act," they wrote. "Soil can also be useful associative evidence in the investigation of wildlife crimes, such as poaching."
At the moment, though, most soil forensic methods are based on scrutinizing the chemical composition of soil rather than its DNA content, the study's authors explained. When genetic analyses are done on soil, they tend to focus on techniques that measure molecular fragment lengths rather than specific sequence compositions, they added, noting that there is "currently no reliable method to compare the DNA content of soils for forensic purposes."
The advent of high-throughput sequencing technologies has raised the possibility of detecting DNA from a wide range of microbial community members in a given soil sample, along with accompanying DNA from other microbes, fungi, plants, insects, and animals.
Such applications are appealing, but may be tricky to pull off without deep sequencing due to excess representation of reads from microbes with the highest representation in common in given soil microbiome.
The limited amount of material available can also complicate matters, the researchers noted, providing an edge to approaches that involve DNA extraction and amplification followed by some form of targeted sequencing rather than shotgun sequencing.
For the new analysis, the researchers used the Ion Torrent PGM to profile DNA isolated from triplicate soil samples collected at two residential parklands from Adelaide, Australia that were roughly three kilometers (nearly two miles) apart.
Linacre noted that the same types of sequence data could be achieved with virtually any of the available next-generation sequencing instruments. For the sake of consistency, though, he and his colleagues decided to stick with the PGM platform, which they'd used for earlier soil sequencing experiments.
Along with libraries prepared directly from microbial DNA extracted, the team prepared DNA from the same samples using whole-genome amplification and AP-PCR prior to sequencing library construction.
Libraries produced with each of the three approaches were each sequenced with different Ion 318 chips by investigators at the Australian Genome Research Facility, who generated an average of almost 673,000 reads per sample on the shotgun sequenced samples, nearly 912,000 reads apiece for samples prepared using whole-genome amplification, and more than 468,000 reads per AP-PCR-prepared sample.
In the case of the samples prepared using AP-PCR, the researchers settled on a primer that targets microbial sequences based on GC-content, Linacre explained.
"We had trialed a few other primers previously and found this sequence to give the most reproducible amplification," he said, noting that the same primer seems to help in generating sequences that can distinguish microbial sequences from a range of soil types, including sandy soils and loamy soil material.
After generating microbial sequences with each of the three methods, the group went on to quality-filter, annotate, and analyze the reads with the help of a server known as Metagenome Rapid Annotation using Sub-system Technology (MG-RAST), a M5NR database meant to determine microbial taxonomy from predicted protein and ribosomal profiles. They also used a SEED sub-systems database designed to offer metabolic insights based on microbial sequence data.
When they compared and contrasted the taxonomic and metabolic information gleaned from sequences produced using each method, the researchers determined that the reads produced with the AP-PCR approach provided the best differentiation between soil samples from two sites.
In contrast, the taxonomic and metabolic profiles generated from the shotgun sequenced and whole-genome amplified samples tended to cluster relatively near one another, making it tricky to tell them apart in some instances.
"[Shotgun]- and [whole-genome amplification]-based metagenomic sequencing approaches showed incorrect and inconsistent discrimination of soil samples according to sampling sites using both taxonomic (protein and ribosomal) and metabolic classifications," the researchers wrote.
In particular, they explained, the profiles produced with these approaches "revealed not only misclassification of the samples between the locations but often between repeat analysis of each sequencing approach."
The team's analysis of randomly sampled subsets of reads generated using each of the three approaches produced the same general taxonomic and metabolic patterns and the full sequence datasets, they noted.
Linacre noted that data from the study has been deposited to the Quantitative Insights into Microbial Ecology, or QIIME, database.
Linacre and his team are primarily interested in pursuing forensics-related applications of the methods described in the study, though the same sorts of data could also offer clues about the biology and ecology of microbes living in soil samples.
"While there are environmental aspects, we were interested in a proof of concept to see which method could give reproducible data from the same location, but separate a relatively nearby location," Linacre said.
In the future, the group hopes to apply this approach to authentic and mock forensic cases in Australia. One of the study's authors, Damian Abarno, is affiliated with Forensic Science South Australia, Linacre explained, noting that the agency is "fully aware of the work and the opportunity to complete validation tests and ultimately apply [it] to casework."
For the current study, the team spent roughly $2,500 generating data for replicate samples at each site. That price tag may be prohibitively expensive for many cases considered in a forensics setting, Linacre noted, suggesting metagenomic soil sequencing may be reserved for select cases for some time, even after the method is further validated and standardized.
"For presentation in a court of law, the development of a sufficient sample size and distinct geographical profiles will need to be bolstered with a determination of the limitations of the method," the team concluded, "including false positive and negative rates."
"This can be achieved via blind trials, mock case work, and a period of casework hardening in order to achieve the levels require[d] for acceptance."