NEW YORK – Short-tandem repeat (STR) testing has been the standard technique for genetic relationship testing for the past few decades, but some scientists believe that SNP arrays and next-generation sequencing could provide a better means for assessing cases in the future.
Driving the shift is a belief that standard STR testing is not always a reliable tool for determining family relationships, which can be complex. While close relationships — such as paternity — are relatively easy to determine using around 20 STR markers, it becomes more complicated to discern the nature of relationships at a greater genetic distance or in contexts where the sample source is highly degraded, such as in human remains of unknown identity.
Recently, a team of researchers from the Center for Human Identification at the University of North Texas Health Science Center in Fort Worth decided to undertake an estimate of how many false positives and negatives might be produced using conventional STR testing in kinship cases.
They determined that for an estimated 10 million kinship tests run globally each year using STR kits, as many as 60,000 might produce incorrect results. The majority of these are related to people being falsely called unrelated, as opposed to calling unrelated individuals close relatives. The results were detailed in PLOS Genetics last month.
"We have been doing missing person identification and relationship tests for civil and criminal cases for more than 20 years," said lead author Jianye Ge, associate director of the center, which provides testing for the US Department of Justice's National Missing and Unidentified Persons System (NamUs) database.
"We process lots of relationship tests and we know what is good, what is bad, and what needs to be improved," said Ge. "We've thought about the question of how many results have been wrong for years, so we finally decided to make an estimate and let the community know that."
As Ge noted, a false result can have a serious impact on people's lives. Such tests are not only used to determine paternity, he noted, but also in migration cases to determine whether immigrants seeking to settle with family members are really genetically related. Different factors can reduce the reliability of a test outcome, from a lab's thresholds for determining what results are inconclusive to basic genotyping errors, as well as false interpretations by technicians.
The risk of a misinterpreted result goes up the greater the genetic distance between those tested, an issue compounded by family relationships that can be hard to discern using STR markers, such as first cousin or half-sibling relationships.
Researchers are now suggesting to take a page from consumer genomics and adopt array-based SNP genotyping or whole-genome sequencing data to more reliably determine relationships.
"Theoretically, with so much SNP data, the accuracy of close relationship testing should approach 100 percent, which could rectify a serious concern of misidentifications, both related and unrelated, and third-degree or more distant relationships may be determined with high accuracy," the authors wrote in their paper.
Ge said it was time forensic labs moved to new technologies. "We encourage the whole community to consider shifting from STRs to SNPs," he said. "That is something we are doing in our lab, where we are validating SNP arrays for relationship testing."
The casework unit within the center is using STRs for relationship testing, he said, while its research unit is validating both Illumina SNP arrays and sequencing panels for relationship testing, particularly missing persons cases. "It will take time for the casework unit to implement these new technologies in real casework," he added.
One outstanding question is why consumer genomics firms, such as 23andMe or AncestryDNA, have been using SNP arrays to determine family relationships as part of genetic genealogy services for a decade or more, but the forensic genetics community is still mulling the transition to new technologies.
Forensic genetics is traditionally more conservative when it comes to technology adoption, Ge noted, and any technology used for legal purposes has to be highly vetted. "Labs need to purchase the instruments, validate the technology, train the people to use it, and have the community accept it, and that takes time," he said.
Another potential roadblock to uptake of new technologies is that the US Federal Bureau of Investigation's Combined DNA Index System, or CODIS, contains STR marker information on roughly 15 million individuals. Whereas consumer genomics companies built their databases from scratch on the basis of innovations from the life sciences industry, the forensic genetics community has both an entrenched method and a large database centered on it. For this reason, labs continue to churn out STRs while adding on arrays or sequencing as supplementary tests.
"In criminal cases, we still need to have the STR results," noted Ge. "CODIS only accepts STRs."
A need for more markers
Forensic genetics may be taking more time to make the jump from STRs to arrays and sequencing than consumer genomics, but such a shift is still inevitable, according to other experts.
David Mittelman, CEO of Houston, Texas-based Othram, which specializes in using next-generation sequencing and genetic genealogy to identify human remains and to solve cold cases, said that the field is certainly trending towards adoption of higher-resolution technologies, which is necessary, as there are specific issues with using STRs that make detecting complex relationships hard.
"Putting aside the technology you use, whether you use sequencing or arrays, in general, it makes more sense to have more markers," said Mittelman. "STRs can only detect a certain amount of distance genetically, they can only detect a close relationship," he said, and will miss relationships between cousins, for example, which SNP arrays or next-generation sequencing can easily determine.
"The power of using anywhere from tens of thousands to hundreds of thousands of markers is that you get a really clear, exact relationship between two people," he noted. "STRs are good if you want to confirm that you are yourself, or someone is your son, but beyond a sibling or a parent relationship, the way you get things done is looking at lots of markers, and because it's now practical to assay lots of markers, it feels almost inexcusable not to move in that direction."
Mittelman, formerly CSO at Family Tree DNA, noted that the reason consumer genomics firms adopted the technology more readily than forensics labs was because they could accept more risk. And since forensic genomics is a more entrenched field and the consequences of making a mistake could be severe, making any kind of technology shift involves far more than buying a new instrument and plugging it in.
"When you are doing work like this, when you are trying to do something that might be used in a court of law or to legally establish your relationship, there is some more vetting that needs to be done," he said. "Whether you are using sequencing or arrays, the technology is totally different. You can't just drop in a new piece of equipment. There are new skill sets required, it's a totally different thing."
Andreas Tillmar, a forensic geneticist at the National Board of Forensic Medicine in Linköping, Sweden, similarly noted that entrenched approaches made the forensics community more wary of adopting SNP arrays and next-generation sequencing. But costs and turnaround time have also made people hesitant to buy a sequencing instrument and use targeted NGS panels.
"The field of forensic genetics, both [for] kinship and criminal cases, is extremely conservative," said Tillmar. "Many of our cases go to court, and in order to prove something in court, you need to have thoroughly validated methods that should be in place for a long time."
Laboratory directors that do wish to switch have to demonstrate improvements in solving cases, as well as a reduction in costs and greater speed, he said, as staff typically needs to be retrained and new IT and laboratory information management systems have to be implemented.
He also noted that, despite some misinterpreted cases, STRs are still informative and their wide adoption over multiple decades has led to a body of data that can be used to interpret results. "We know a lot about these markers, and for SNPs, we don't have that in-depth analysis yet at the same level," said Tillmar. "We started with STRs 20 or 30 years ago and it's hard to leave that behind."
The National Board of Forensic Medicine in Sweden is a governmental authority that handles all paternity and kinship cases in the country, roughly 10,000 per year, Tillmar said. In addition to standard paternity tests, it also carries out kinship tests related to immigration cases, as well as a few hundred missing person identification cases each year.
For the past five years, it has used massively parallel sequencing as a supplementary test to STRs in cases with unclear outcomes. "If we do not obtain the necessary likelihood ratio, we add on supplementary SNPs," he said. "It's just a couple hundred SNPs, not these high-density panels, but it still helps us solve all the cases we have."
Adding sequencing, even as a supplementary test, is still a challenge, though. "It is doable to implement it in a kinship lab, but you need to have individuals who can implement it in the lab and educate the other staff," he said. "But thanks to sequencing, we have solved cases that we cannot solve using STRs."
Ranges of application
According to Colleen Fitzpatrick, president and founder of IdentiFinders International, a Huntington Beach, California-based forensic genealogy service provider, it is targeted SNP panels run on next-generation sequencing instruments that are more likely to become an approach of choice as the forensics community transitions to using new technologies alongside STRs.
"All technologies have their strengths, their ranges of application," said Fitzpatrick, a pioneer of forensic genetics and the author of the 2005 book Forensic Genealogy, widely considered a touchstone for the industry. "If I was doing a fresh sexual assault case, I would probably use a microarray, because you have fresh DNA, the quality will be great, and so it's less of a risk," she said. But in other cases, particularly for identifying degraded remains, sequencing is the preferable option. "Arrays are not good for highly degraded DNA," she said. "It's like trying to read a damaged book."
She noted that forensic genomics firms, such as San Diego-based Verogen, have been developing targeted panels that enable the use of next-generation sequencing for forensic genealogy. These panels offer a limited number of SNPs, similar to the approach Tillmar's lab is using in Sweden. This "right down the middle" approach gives users the capabilities of sequencing, plus the "leanness and meanness of microarrays," said Fitzpatrick.
"What I see in the future is [targeted] sequencing replacing whole-genome sequencing and microarrays in the crime lab," she said, noting that neither technology would replace the STRs that form the backbone of the CODIS database, but that targeted NGS would be implemented alongside STR testing.
"CODIS will always be with us, there are 15 million people in that database, and we can't go back and redo all 15 million," said Fitzpatrick. "We are stuck with STRs, we are stuck with CODIS, but it will be accompanied by NGS in parallel."
She also noted that while it seems that the field has been slower to shift technologies than, say, the consumer market, arrays and sequencing actually have been used to solve cases over the past decade. "There is a lot of prior art," she said. "This has been going on since at least 2011." She noted, however, that the success of consumer genomics companies has put pressure on the forensic genetics community to make a similar jump. While STRs came out of the forensics community, SNP testing emerged from the biomedical industry, which was rocketing ahead with innovation in the 2000s and 2010s while forensic geneticists continued to use STRs.
But the emergence of GEDmatch, a free online database that allows customers to upload, compare, and analyze their autosomal DNA data from consumer genomics firms, changed that.
"GEDmatch connected the forensic world with the direct-to-consumer world, and that is when the revolution took off," said Fitzpatrick. "You had the commercially-driven side and medical research-driven side, and then the forensics side, which was moving slower," she said. "You had to connect them, and the connection was GEDmatch."
Fitzpatrick noted the resolution of the Golden State Killer case in 2018 as being a particular watershed for the community. Despite that success, though, she noted that genetics is only one piece of the puzzle when it comes to resolving criminal cases. It might form the basis for an investigation, but other evidence has to be gathered before any case can be made. So even though technology has to be thoroughly validated before being implemented, the future of a criminal case doesn't ride alone on a genetic result. "Whatever we do in genetic genealogy, it is still a lead," she said. "It doesn't have to stand alone in court."