This article was originally published June 24.
A research group from the University of North Texas Health Science Center has tested Life Technologies' Ion AmpliSeq Human Identification SNP panel for use in forensics, comparing it to a SNP panel run on the Illumina Genome Analyzer.
Reporting their results this month in the International Journal for Legal Medicine, the team found that the two platforms were concordant on 94 out of 95 SNPs. Additionally, they tested a range of starting input from 10 nanograms of DNA down to 100 picograms of DNA to see how the machine performed on sample sizes that would be found at crime scenes.
Jianye Ge, an author of the study and assistant professor within the Institute of Applied Genetics at North Texas, told In Sequence that the group is currently evaluating both the PGM and Illumina MiSeq for use in forensics due to the systems' abilities to generate much more information compared to the standard short tandem repeat, or STR, analysis with Sanger sequencing.
Looking at SNP panels, said Ge, will enable researchers to identify much more information from a sample, including population and physical features. And, it will be more amenable to highly degraded samples, which are often found in crime scenes.
"Forensic DNA databases can be expanded from [the current 13] STR markers to hundreds of STR markers and hundreds of SNPs," he said. More markers will provide higher accuracy, in terms of identifying the individual sample and relationships between samples, he said.
Additionally, using next-gen sequencing in forensics could "make it easier to get more knowledge about the samples collected from crime scenes."
In the most recent study, the team evaluated the Ion AmpliSeq HID kit, a panel composed of 103 autosomal SNPs and 33 Y-chromosome SNPs, using the Ion Torrent PGM 314 chip. They tested samples from four individuals and compared the results to a different SNP panel on the Illumina GA that shared 95 SNPs with the AmpliSeq panel.
Using 10 nanograms of DNA, the researchers identified all 136 SNPs in the AmpliSeq panel. Similarly, with just one nanogram of DNA, all 136 SNPs were detected, but one locus showed extreme heterozygote imbalance on allele coverage. When they reduced the input to 100 picograms, however, an average of 1.6 SNP loci per sample were not detected, and an average of 4.3 SNPs showed heterozygote imbalance.
Compared to the GA panel, all SNPs were concordant except for one. The discordant SNP was caused by a PGM sequencing error due to the SNP's location in a homopolymer region.
"It's a problem of both the chemistry and the software," said Ge of the missed SNP in the homopolymer region. "We also need a better definition of the threshold to determine what is an error versus a true SNP."
For the technology to be used in forensic applications, a certain amount of error may be okay for some research purposes, he said, but if "you want to use that data to convict people, you have to have very high accuracy."
Next, the team evaluated the performance of the AmpliSeq panel on increasingly lower amounts of starting DNA. At 10 nanograms, "there is consistently high coverage with little variation between samples," the authors wrote. There was, however, coverage variation among the SNPs themselves, with each SNP in general displaying similar coverage from sample to sample. The lowest coverage was a SNP on the Y chromosome, which was only covered between five- and nine-fold among the three male samples. The authors hypothesized that coverage differences are likely due to PCR amplification and that modifying the primers or primer concentrations could result in more balanced coverage.
Reducing input DNA from 10 nanograms to one nanogram did not have a big impact on the results. Coverage of SNPs was the same for the one female sample and slightly lower for the three male samples.
Additionally, all but one SNP genotype was detected with one nanogram of DNA. When the starting material was reduced further to 100 picograms, however, up to six SNPs were missed in some samples. The Y chromosome SNPs were more likely than the autosomal SNPs to be missed.
"The technology is very promising," said Ge, but to be used in forensics, "it will need to be developed more," he said.
Being able to work with limited amounts of DNA will be critical, he said. "In many forensics cases you may only have 500 picograms or 100 picograms of DNA," he said.
Nevertheless, he said that he thinks next-gen sequencing offers a number of benefits over the conventional STR typing. Currently, his group is evaluating both the PGM and MiSeq, and is testing STR profiling, SNP panels, and targeted mitochondrial DNA sequencing.
He said the group has a forthcoming publication evaluating a SNP panel on the MiSeq.
One major application of sequencing could be in parsing mixed samples, he said. For example, if two people drink out of one bottle of water, "when you swab the bottle you get DNA from both and it's hard to tell which is which," he said.
He said that his group is also collaborating with Life Tech and Illumina to develop better software for calling the alleles. For instance, he said, errors are often introduced because most analysis pipelines first align the reads to a reference genome before calling variants.
Ge's group at North Texas is not the only one working on developing sequencing for forensic uses. Researchers at Yale University, led by Kenneth Kidd, collaborated with Life Tech to develop the AmpliSeq HID panel, and Kidd is also working on an ancestry SNP panel and a phenotype panel (IS 10/23/2012).
In addition, Mark Wilson, director of the forensic science program at Western Carolina University, is doing mitochondrial sequencing on the MiSeq to detect rare variants below 1 percent frequency. Illumina does not currently offer a specific forensics kit, but includes forensics genomics on its website as an application of MiSeq.
Ge anticipates that if sequencing technology becomes "more reliable with higher accuracy," it could be used in forensic casework within the next five years.