This article was originally published Sept. 8.
NEW YORK (GenomeWeb) – A group of researchers at Johns Hopkins University has designed a next-generation sequencing-based method that analyzes blocks of closely spaced SNPs that could be used in human identity testing for forensics applications; limb, organ, or bone marrow transplantation; and paternity testing.
The researchers demonstrated a proof of principle for the method in a study published last month in the Journal of Molecular Diagnostics. They designed an NGS-based assay on Thermo Fisher's Ion Torrent PGM to target a haplotype block of the HLA-A locus and demonstrated that the assay could identify DNA from one individual from a mixture of DNA with a lower limit of detection of 0.01 percent.
The method could also be applied to other haplotype loci, the authors wrote, and selecting the appropriate loci "may be influenced by the patient's ethnic background, number of discriminating SNPs, and ease of primer placement."
James Eshleman, associate director of the molecular pathology lab at Johns Hopkins, told In Sequence that the goal was to design an assay more sensitive than current methods that rely on short tandem repeat analysis by capillary electrophoresis, but able to overcome the limitations of next-gen sequencing's error rates.
Currently, forensics analysis and applications like bone marrow engraftment monitoring rely on STR analysis because STRs are highly variable between individuals and occur throughout the human genome. Typically, laboratories use multiplex PCR-based kits and CE sequencing for STR analysis, but the method has a lower limit of detection of only 3 percent to 5 percent.
Evaluating SNPs instead of STRs could theoretically be a more sensitive approach, Eshleman said, as long as enough SNPs are evaluated such that a given individual yields a different result from a mixed sample and as long as there is a way to get around the error rates of next-gen sequencing systems.
To tackle both of these problems, his team identified a stretch of DNA containing enough SNPs such that there is a small likelihood of any two individuals containing the same bases at all of the SNP positions. For the initial test, the group chose a haplotype block on the HLA-A gene containing 18 SNPs surrounded by non-polymorphic DNA; however, he said, other suitable haplotype blocks exist throughout the genome and could be used to design assays.
Next, they analyzed two cell lines with known HLA-A genotypes to establish accuracy and the lower limit of detection. The researchers chose cell lines whose two alleles of interest varied from one another by 11 SNPs, and each varied from the commonly shared allele by seven SNPs. They made cell mixes varying from one in 1 million to one in 100 using a total of 10 million cells for each dilution. They then isolated and PCR amplified 600 ng of DNA and sequenced each sample twice.
The assay was highly precise at the 0.1 percent cell mix with a coefficient of variation of just 0.12 percent. At the 0.01 percent mix, the coefficient of variation was higher, at 0.4 percent. Nonetheless, the team was able to determine the minor HLA-A allele frequency in all four replicates at that level. Decreasing the cell mix even further to 0.003 percent, they were only able to identify the minor allele frequency in three of four, so established the lower limit of detection at 0.01 percent, although Eshleman said that is a conservative limit, and the assay could likely have an even lower limit of detection.
The group then tested the assay on 18 samples from patients that had received bone marrow transplants and whose original and donor HLA genotypes were known and varied by at least four SNPs. With STR analysis the samples tested as all donor DNA, but using the NGS-based SNP assay, all but one sample tested positive for patient DNA with samples having between 0.001 percent and 1.47 percent patient DNA — below the lower limit of detection of STR analysis.
Other researchers have also looked at the feasibility of using NGS-based approaches in forensics, rather than STRs, due to sequencing's increased sensitivity.
Researchers at Yale University described a similar method in a study published in Forensic Science International: Genetics that demonstrated that sequencing microhaplotype loci could identify individuals and also determine their ancestry and relatedness to others.
Another group from Eurofins in Germany found that using SNPs, rather than STRs, could tell the difference between identical twins. That group's approach, however, relied on whole-genome sequencing rather than targeted sequencing.
In addition, a team from the University of North Texas has been testing SNP panels on both the PGM and Illumina's MiSeq system for use in forensics.
Eshleman said that aside from the HLA-A locus, his group has identified other haplotype regions throughout the human genome on which similar SNP-based assays could be developed. They used information from the 1,000 Genomes Project to search for regions no longer than 300 bp that contained at least nine SNPs from the three populations — European, Asian, African — that were surrounded by at least 20 bp of nonpolymorphic DNA and identified 4,349 loci acorss the genome. Eshleman said that the team is now working to validate some of those loci patient cohorts.
"Selection of suitable loci may be influenced by the patient's ethnic background, number of discriminating SNPs, and ease of primer placement," the authors wrote.
Moving forward, Eshleman said that the assay could have applications in forensics, transplantation, and even monitoring residual disease. "We're currently applying for funding to further develop this as a full-blown assay," he said.