Researchers in the US and Europe are working to move next-generation sequencing into forensic applications, which could increase discriminatory power, as well as enable previously untestable samples to be analyzed.
Currently, forensic profiling is based on STR analysis with capillary electrophoresis, or analysis of mitochondrial DNA with PCR and Sanger sequencing.
But next-gen sequencing opens up new possibilities. Researchers are now looking at moving both STR profiling and mitochondrial DNA analysis to next-gen sequencing, but are also looking to develop SNP-based targeted sequencing panels for forensics, which would enable researchers to identify a person's ancestry, hair color, or other defining characteristics based on their DNA.
Additionally, next-gen sequencing "allows us to look at all those markers in one method," Walther Parson, a professor at the Institute of Legal Medicine at Innsbruck Medical University in Austria who is developing forensic panels on Life Technologies' Ion Torrent PGM, told In Sequence.
While most researchers are looking to first develop targeted sequencing based panels, others, such as Yaniv Erlich at the Whitehead Institute, are developing methods for STR profiling from whole-genome sequencing data (IS 5/1/2012).
Innsbruck's Parson is developing two panels to evaluate mitochondrial DNA using Ion's AmpliSeq technology and the 314 and 316 chips. One panel is for analyzing the full mitochondrial genome, which would be ideal for high-quality samples where DNA is not limited, as in a blood sample; and the second panel will only analyze a portion of the mitochondrial genome for samples with limited or low-quality DNA, which are most likely to be found at a crime scene.
"We can't do a full [mitochondrial] genome from a case sample," he said, because it "involves amplification of relatively large fragments that are not found in those case work samples."
Instead, he is working on identifying targets within the mitochondrial genome that "have a high discrimination power and other parts that give us information about the phylogenetic background of the sample."
Parson said his team is currently deciding how large the targeted mitochondrial DNA panel will be.
"It depends on primer binding sites and how they amplify with degraded DNA, and also there is some limitation to the number of loci that are available in a degraded sample to be amplified with a multiplex reaction," he said.
He expects to have initial panels available this year, but "to be able to produce a product that is robust and can be validated in the forensics context" will take longer, he said — at least a couple of quarters into 2013.
The advantage of using next-gen sequencing is that many different types of markers can be combined into one panel, he said.
For instance, in the longer term, his team plans to combine mitochondrial DNA markers with nuclear DNA markers in one panel. Currently, researchers use different markers for different applications, like paternity testing or case work, he said.
"The PGM allows us to look at all those markers in one [panel]," he said. "That is an advantage that traditional methods don't have."
Additionally, given that a forensic sample is typically "precious," such a panel would allow researchers to "get as much information as possible out of that sample."
Mitch Holland, director of forensic science at Pennsylvania State University, is also looking to use next-gen sequencing of mitochondrial DNA in forensic applications. Last year, his team published a study demonstrating the potential of Roche's 454 GS Junior to analyze the hypervariable region 1 of the mitochondrial genome.
Since then, he has expanded this work to include the hypervariable region 2 and is now also testing the Illumina MiSeq platform and plans to publish a study comparing the two platforms.
Holland explained that next-gen sequencing offers better discriminatory power than Sanger sequencing because it is able detect minor variants. This not only increases the sensitivity of the assay, but also enables samples to be analyzed that were previously too convoluted. Samples that contain mixtures of DNA can often not be untangled via Sanger sequencing, but it would be possible to analyze them with next-gen sequencing, he said.
In last year's study, Holland's team evaluated 25 different samples by both Sanger and 454 sequencing. Both methods were able to detect a heteroplasmic variant in only one of the samples, which had a heteroplasmic variant present at around a 20 percent frequency. However, 454 sequencing was able to detect heteroplasmic variants that were below the limit of detection for Sanger sequencing in an additional 10 samples. In these samples, the minor variant was present at a level of between 0.33 percent and 4.5 percent.
Holland's team also demonstrated that 454 could both detect heteroplasmy and also deconvolute mixtures of mitochondrial DNA.
Despite obtaining good results with the GS Junior platform, Holland said the platform did have some issues with homopolymer regions, so the team is now evaluating the MiSeq.
"We initially chose the 454 platform because of the length of reads," he told In Sequence. But, he added, "while the 454 produces reliable data, it does have an issue with homopolymers, and there are quite a few of those in the mitochondrial genome."
Holland added that for now, his team is sticking with analyzing targeted regions of the mitochondrial genome, such as HV1 and HV2, because analyzing the entire mitochondrial genome is often not practical with case samples.
"A small hair shaft fragment that's a couple of centimeters long, you're not going to get a whole genome," he said.
Meanwhile, another group of researchers led by Kenneth Kidd at Yale University is validating a set of SNPs to be used on an upcoming Ion Torrent panel for human identification.
Life Technologies is currently offering a version of the panel through its Ion Community website for early access users that includes around 130 SNPs, including 92 that Kidd's team published in Human Genetics in 2010.
The company plans to release a commercial kit next year that improves on this initial panel.
Additional SNPs, such as ones on the Y chromosome, and others were added to the final panel, "so that if one tests for all 130 SNPs, you've got an astronomically small probability of anybody else having, or ever having had, that specific genotype," Kidd told In Sequence.
The panel is also valid for any population in the world, he said, "so it becomes an extremely powerful tool — much more so than the current forensic markers," which have different frequencies in different populations.
Aside from the individual identification panel, Kidd is also working with Ion Torrent on an ancestry SNP panel and a phenotype SNP panel. The ancestry panel will be based on a study his group published in 2011 in Investigative Genetics that validated 128 SNPs in more than 4,800 individuals from 119 population samples.
Kidd said that he is still improving this set of ancestry SNPs. One application for the ancestry panel would be in cases that rely on missing person lists to help identify human remains. For example, if there is a list of 1,000 missing persons, and the ancestry panel can say the human remains were of East Asian descent, there may be only a handful of individuals on the list with that background. "That's a big help in trying to identify who that body was," he said.
The phenotype panel will contain SNPs that are known for skin, eye, and hair color, Kidd said. Additionally, other researchers are trying to identify SNPs that correlate with facial structure, he said.
Next-Gen on Trial
DNA evidence obtained from next-gen sequencing has yet to be used in a courtroom, but Penn State's Holland said he thinks the field is moving quickly in this direction. He thinks the first application to see the courtroom will be in mitochondrial DNA sequencing.
"Historically, mtDNA sequencing has been done in forensics with Sanger sequencing," he said. "The community is familiar with sequencing, so moving to next-gen won't be as big a leap."
On the other hand, STR profiling evaluates the length of the STRs, so moving to next-gen introduces a new way of analyzing those markers. And the SNP-based panels provide both a new technology and new markers that have not yet been used in the courts, he said.
Holland anticipates that NGS-based mitochondrial sequencing could be used in the courts in as soon as two years.
Kidd also thinks it's only a matter of time before next-gen sequencing is introduced into the court system. For data generated by next-gen sequencing to be admissible in court, it would have to pass both what is known as the "Frye standard" and the "Daubert principle," rulings that stipulate that the method should be accepted by the scientific community and be based on strong scientific principles.
Whether individual forensic laboratories are able to adopt the technology is another question. Holland said that he has seen an increased interest in the community. A cost analysis still needs to be done, but he thinks mtDNA testing with NGS can be done at comparable costs, if not cheaper, than current Sanger-based testing methods.
Additionally, sequencing costs continue to fall, and with the availability of benchtop sequencers, capital costs are now low enough that many labs can afford machines.
The main hurdle to adoption, Holland anticipated, will be data storage. Next-gen sequencing "generates enormous amounts of data that [forensic scientists] aren't used to seeing," he said. And "because it's a forensic environment, you can't just throw things away."
As a result, figuring out cost-effective and secure ways to store and manage data will be critical, he said.