Name: Arthur Eisenberg
Title: Professor and Chairman of the Department of Forensic and Investigative Genetics and Co-Director of the Center for Human Identification, University of North Texas Health Science Center
Experience: University of North Texas Health Science Center since 1989, full professor since 2004, chairman of the Department of Forensic and Investigative Genetics since 2009;
Chairman of the United States DNA Advisory Board, 1998.
Education: PhD in molecular biology, MS in molecular biology, and BS in molecular biology from the State University of New York, Albany
Since Arthur Eisenberg helped establish the first DNA paternity and forensics laboratory in the world in 1984, he has seen forensics technology evolve from restriction fragment length polymorphism, or RFLP, testing to PCR-based testing and Sanger sequencing — and, more recently, next-generation sequencing.
Eisenberg chairs the Department of Forensic and Investigative Genetics at the University of North Texas Health Science Center. The university's Institute of Applied Genetics recently purchased an Ion Torrent PGM, with intentions to purchase a second PGM instrument, and plans to develop protocols on the system for forensic testing.
Eisenberg thinks that next-generation sequencing can eventually replace the PCR and Sanger-based methods that are currently used for forensics, and foresees a day when all types of forensic testing — STR profiling, mitochondrial sequencing, and SNP testing — can be done on a single platform.
Eisenberg recently spoke with In Sequence about the different applications for which the PGM could be used, and the validation required to use the technology in court.
Can you provide a little background about forensic testing and the methods currently used by your lab and others?
Most DNA testing is based on STR — short tandem repeat — markers. And the STR systems are run on capillary electrophoresis, and we look at allele variations based upon length. So, that's the predominant form worldwide.
There are probably on the order of 38 million DNA profiles that are available for associations with forensic crime. In the US, our national databases have about 11 million or so DNA profiles from individuals who have been convicted of crimes. There are several hundred thousand forensic evidentiary profiles. About half the states allow the collection of DNA upon arrest.
The other national databases are those involving the missing and unidentified, and our lab is the largest contributor to those databases. Basically, our lab and the [US Federal Bureau of Investigation] do the majority of the identification of human remains and missing persons. For those types of analyses, in addition to STRs, we do mitochondrial sequencing using conventional dideoxy sequencing. And we're also evolving into [evaluating] single nucleotide polymorphisms.
We see the PGM and next-generation sequencing as sort of the evolution, in that, simultaneously, we can do all those things. We can do STRs, we can certainly do mitochondrial sequence analysis, and we can analyze for large numbers of single nucleotide polymorphisms. Rather than looking at any individual systems, be it Y-STRs, be it autosomal STRs, be it mito-sequencing, be it SNPs, we can do everything on next-generation sequencing platforms.
Why did you decide to purchase the PGM to use in forensics as opposed to another next-generation platform?
The instrumentation certainly fits within the budgetary realm of conventional CE instruments. The platform and the up-front library construction are on the order of $100,000 to $120,000. Forensic labs are used to paying $150,000 for capillary electrophoresis instruments. So it's not the same price range as the Illumina instruments, the 454, SOLiD, the Proton. We don't need to sequence the whole human genome. We can simultaneously look at several hundred different markers on a very cost-effective platform.
What's your timeline for implementing the PGM in forensic testing, and what applications will you first transfer over to the PGM?
We're just starting. We have the instrument installed. We're going to first start off with mito-sequencing.
The forensic community tends to look at two regions … called hypervariable region 1 and hypervariable region 2. In essence, we sequence about 750 to 800 bases, but there's a lot more information that can be used for distinguishing maternal lineage.
Right away we can implement sequencing on [the PGM] versus traditional sequencing and get a lot more information in a relatively cost-effective manner.
Then, when you add barcoding in, it becomes highly cost effective. With traditional sequencing, there's not an economy of scale of incrementally sequencing one, five, 10, 15 individuals, but with barcoding you could simultaneously sequence multiple individuals and it becomes very cost effective.
So the first implementation would be to transition mitochondrial DNA analysis over to sequencing on the PGM. And we expect to be able to do that well within six months to a year. Probably more on the time frame of six months.
What about STR profiling? Is that something you plan to do on the PGM?
It's theoretically possible to do STRs on [the PGM]. There are so many profiles in our national databases, and STRs are not being replaced and are not going to be replaced for a long time. That's the bread and butter of DNA forensic labs. There are probably over 180 DNA forensic labs in the United States alone. The largest database is in China. They probably have about 14 million people typed.
These databases are growing very rapidly with STRs. So, in order to be of utility, the PGM or any next-gen sequencing instrument is going to have to have the ability to do STRs.
Clearly the most abundant type of polymorphisms are SNPs and we'll see a transition, but that will take time, maybe five to ten years. And the PGM is sort of a bridge between the old technologies, STR technologies, and the newer technologies, sequence-based and SNP technologies. We see the PGM being part of this evolution that will ultimately end up being SNPs, but it's going to take a number of years.
So far, there has been limited work done on these types of instruments for STR profiling. Because of the size of the STRs, the read lengths are going to be very important. But that's one thing where the PGM shows a lot of promise. The read lengths are now about 200 base pairs, and in the not too distant future they'll be 400 bases. Certainly at 200 bases we have the possibility of doing quite a few [STR markers]. As the reads progress to 400, we should be able to cover all of the STRs that are currently used in forensics.
The beauty of the PGM is that it will allow us to do conventional STRs and then help us in the transition into sequencing technologies and into single nucleotide polymorphisms. And we could do hundreds of these markers simultaneously.
What about longer-term goals, or new applications? Will the PGM, or next-gen sequencing in general, enable anything that is difficult to do using Sanger sequencing or PCR?
The single largest variation between individuals is SNPs. There have been many SNPs that have been developed for the field of identity testing and we can use SNPs to do ancestry testing.
One of the things that we're really interested in, being one of the largest labs in the country and the world that works on human remains, is there are SNPs that are associated with phenotypic features. One of the programs that we have oversight over is called NamUs — the National Missing and Unidentified Persons System.
There are many different websites that people could go to, to look to see if their loved one has been identified. Well, NamUs is an attempt by the Department of Justice to consolidate those places. NamUs consists of two databases. One database is where families list their loved one as missing. They can put up pictures and so forth. The other database is where medical examiners, coroners, law enforcement [officials] can list physical features or perhaps facial reconstructions on unidentified bodies. The idea is that people can look at both of those databases.
Well, there's nothing as powerful as a good facial reconstruction. We have forensic artists that work with us and if we can provide them with additional information like hair color, eye color, facial morphology — there are many different genetic markers in terms of phenotype that are being discovered in our genome. [We want to] use the Ion Torrent technology to try and get genetic information on some of these phenotypic markers that will help provide a more life-like representation of what the individual might have looked like when they were alive.
There are any number of these SNP markers that are being discovered, and the PGM offers a tremendous opportunity to simultaneously look at many of these markers and then provide genetic information to improve the quality of facial reconstruction.
That application is certainly going to take a little longer — over the next couple of years. The community needs to streamline the process of constructing the libraries, reduce the number of manipulations, and turn [the PGM] into a real forensic tool. This is something that's going to take a few years.
Certainly, the first application is mitochondrial sequencing. That's a no brainer. We can do that now. We can apply it to forensic applications where we may have more challenging samples, like limited quantity. For instance, the most common crime scene sample is shed hair. If you pluck a hair, you have the root, and in the root are the cells that secrete the hair. However, in a shed hair you have no root, you have no cells. But as that hair grows, cytoplasm is excreted into the shaft of the hair and within the cytoplasm are mitochondria. So the only type of forensic DNA analysis you can do on shed hair is mitochondrial DNA analysis.
So this may open up a new avenue in terms of collecting forensic evidence. There are only a handful of labs in the country that do mitochondrial DNA sequencing for forensic casework or for missing persons. So [next-gen sequencing] technology could make it possible for other labs to do this type of analysis.
At any crime scene there is an abundance of hair, and you have to be able to put the presence of that hair within a forensic format. For instance, you could have people living in a house where a crime was committed. If hair is found that's not associated with the people living there, it could be an investigative lead. The technology opens up a whole realm of possibilities that couldn't be done using conventional methodologies.
What do you have to do in terms of validating the system or developing protocols?
Ultimately, we need to introduce the techniques in courts, so there's very stringent validation work that needs to be done. We're taking it well beyond the traditional research environment through very extensive validation. One of the things that people here have been involved with, myself included, is that a number of us were on the DNA advisory board to recommend to the director of the FBI national standards. As part of our national standards, there are any number of specific experiments that must be done in order to demonstrate the reliability and reproducibility of the data we obtain.
This is new ground. Although we've been using sequencing technology for many years, probably a dozen or so years for mitochondrial DNA, we've never done it on next-generation platforms for the purposes of forensic casework. So we have to do very extensive validation studies to demonstrate to the court that this technology is reliable and reproducible. Part of our initial work is to do those necessary validation experiments and to publish them and to be able to defend them in the legal system.
For clinical labs, there is CLIA certification and CAP accreditation. Is there something similar for forensic validation?
Oh sure, there's ISO accreditation. Several of us have worked on the College of American Pathology relationship testing and forensic committees. So we participate in those types of quality control testing. Several times a year we have to do proficiency testing.
Our lab is ISO accredited. There are two major accrediting agencies in the forensics field — the [American Society of Crime Laboratory Directors, Laboratory Accreditation Board] and Forensic Quality Services International.
We are also accredited by the American Association of Blood Banks for relationship/parentage testing. So the forensic field has the same type of accreditation process and proficiency process. We are audited every year. Every other year, it's external. So we have the same types of demonstrated requirements for reliability as in the clinical lab environment.