NEW YORK (GenomeWeb News) – The story of modern molecular forensics begins, unlikely enough, with a hunk of seal meat.
In 1984, University of Leicester professor Alec Jeffreys, the inventor of DNA fingerprinting, was at the British Antarctic Survey in Cambridge working to isolate the gene for myoglobin from a grey seal specimen. His plan was to use the seal gene – which is highly expressed – as a probe for isolating the less abundant human version.
Upon isolating the seal myoglobin and, subsequently, the human myoglobin genes, Jeffreys found that the latter contained a short tandem repeat, a repeating sequence of two-to-six base pairs of DNA. STRs were another of Jeffreys' research interests; he and others considered them potentially promising genetic markers. Researchers had had limited success, though, in identifying such sequences in humans.
A number of the STRs that had been previously isolated, however, featured sequences similar to the myoglobin STR, which gave Jeffreys the idea of using it as a probe to discover new ones. Hybridizing the myoglobin STR to a blot containing DNA samples from several different people, Jeffreys found that he could distinguish between the samples based on the probe's binding. Because the lengths of the STR loci were highly variable between the subjects, they appeared at different points on the blot, creating patterns specific to each individual.
Almost two decades later, comparison of STRs remains the primary tool of molecular forensics. The technique was first applied to criminal investigation in a 1986 double murder case in which Jeffreys used STR matching to initially rule out a false confession and then ultimately identify and convict the actual killer. Since then, DNA profiling has become standard practice worldwide, and in the US has been used in more than 200,000 criminal investigations since the year 2000.
In its early incarnations, DNA fingerprinting used restriction fragment length polymorphism analysis, in which loci were compared by extracting them using restricting enzymes and separating the resulting fragments using gel electrophoresis. This process, however, was cumbersome and time consuming, making it poorly suited to widespread use.
RFLP analysis "was very important for a time," said Lawrence Kobilinsky, chair of the department of sciences at New York City's John Jay College of Criminal Justice. "But the problem is that it's a long, drawn out process – it takes about two months to get a result. And it's very messy. So it's not a good technique when you have a lot of cases to deal with."
And so, polymerase chain reaction came to take its place. Invented a year before Jeffreys' initial discovery of DNA fingerprinting, PCR, Kobilinsky said, proved " a fabulous tool" for forensics, allowing for the amplification of "even the smallest amounts of evidence." Instead of isolating loci of interest via restriction enzymes, technicians could now simply amplify them using loci-specific primers and detect them via electrophoresis.
Today, essentially all crime labs use PCR combined with capillary electrophoresis for DNA analysis, said John Butler, special assistant to the director for forensic science at the National Institute of Standards and Technology. And, he noted, this uniformity of technique extends to the platforms used, as well. The DNA fingerprinting business is currently dominated by two companies – Life Technologies and Promega – with labs typically using Life Tech's Applied Biosystems 3100 and 3500 Genetic Analyzer instruments and either its AmpFLSTR Identifiler PCR kit or Promega's PowerPlex product.
Slow adoption for new technologies
As is perhaps to be expected given the stakes, criminal forensics is slow to adopt new technologies, said Mechthild Prinz, an associate professor at John Jay. Formerly director of the Department of Forensic Biology in New York City's Office of Chief Medical Examiner, Prinz noted that while techniques like DNA microarray analysis and next-generation sequencing are being investigated by forensics researchers, they have yet to make it into a courtroom.
"In court for these admissibility hearings, you need a very large body of work proving that your method really works." she said. "You need peer reviewed publications; you need to have presented it at meetings; you need other experts to agree with you."
She offered one particularly outré example of the sort of "evidence" such standards are meant to keep out: "Once for the [National Institute of Justice], I had to look at a proposal of someone who claimed he could literally develop a photographic picture [from the brain of a victim] of the last image they had seen before they died."
The Federal Bureau of Investigation maintains the Combined DNA Index System, or CODIS, a national database that currently contains DNA samples from roughly 12 million individuals. That database is fed with results from some 200 state and local laboratories across the country, each of which must be FBI-accredited. In 28 states, DNA samples are collected when a person is arrested for a felony. All 50 states collect DNA from individuals upon conviction of a felony.
Over the past several years, a number of states have begun allowing DNA evidence based upon familial searches, in which suspects not in the CODIS database are identified via familial links to relatives who have been registered in the database. Such searches use similarities between two individuals' STRs to determine the likelihood that they are related, the notion being that relatives will have more alike DNA profiles than would random members of a population. In the case of male suspects, STRs from the Y chromosome can also be compared to further establish relationships between fathers and sons or brothers.
Among the most prominent of these cases was California's investigation into a serial killer dubbed the "Grim Sleeper," believed to be responsible for at least 10 murders in the Los Angeles area since 1985. Police had managed to recover DNA from a number of crime scenes attributed to the killer, but had been unable to match these samples to any in the CODIS database. A familial search, however, was able, via analysis of Y chromosome STRs, to establish a patrilineal relationship between the Grim Sleeper's DNA and that of a man who had had his DNA collected upon being convicted of felony weapons charges. Based on the presumed age of the killer, investigators identified the man's father, David Franklin Jr., as a likely suspect. They then surreptitiously collected Franklin's DNA from an unfinished piece of pizza and matched it using standard STR analysis to the DNA collected at the murder scenes.
Franklin was arrested in 2010 but his case has yet to go to trial. In the meantime, however, another California case based on familial searching has made it to court. Last month, Elvis Garcia, who police identified using the technique, was convicted to 65 years in prison for the 2008 rape and robbery of a barista at a Santa Cruz coffee shop.
According to Gary Sims, case-work laboratory manager at the Jan Bashinski DNA Laboratory in Richmond, Calif., which handled the Franklin and Cruz cases, his facility has performed around 70 familial searches since the state approved the technique in 2008 and currently does approximately two a month. One potential limitation of the method is the possibility of false positives, given that such searches return not exact matches but rather ranked lists of the likeliest matches based on the STR analysis.
Sims said that based on his lab's use of both the traditional STR loci and the Y chromosome STRs, "we have the expectation that if there is a first-order relative in [the database], we are more likely than not to detect them." However, he noted, such searches are difficult to perform at a national level due to differences in the two STR multiplexes — Life Tech's AmpFLSTR Identifiler PCR kit and Promega's PowerPlex — commonly used.
The FBI requires that every sample deposited in the CODIS database be tested for the same 13 STRs, and, in fact, the Life Tech and Promega products actually test 15 STRs. The extra two STRs differ in each kit, however, meaning, Sims said, "that if you were to compare [for instance] samples from California with samples in Virginia, they would share those 13 [loci] in common, but they wouldn't share the other two." One factor that could aid such searches is the impending move of US labs to new Life Tech and Promega kits testing 24 different loci.
According to NIST's Butler, this shift is expected to take place within the next few years.
The extra statistical power provided by the additional loci "will help hugely," Sims said, adding that it would prove particularly useful in the case of female candidates where Y chromosome analysis isn't an option. "Right now around 10 percent of our database is female, and we really don't have a good way to address that."
Mitochondrial DNA, which is passed matrilinealy – offers one potential option, Sims noted, but, he said, in practice, the amount of mitochondrial DNA present at a crime scene is typically very small, and often drowned out by DNA from other sources.
"There are limited cases where mitochondrial DNA might work," he said. "But really having additional autosomal work is going to be a big boon for this."
Beyond that, the expanded panels will also help standardize DNA profiling globally, said Sims' Bashinski lab colleague Steven Myers, senior criminologist with the California Department of Justice.
"The real beauty of the new kits is that they will help us with investigations across international lines, because they're meant to overlap with more agencies' required loci," he said. Different countries, he explained, often prefer to look at different loci, as some STRs are particularly useful for discriminating between suspects in certain ethnic populations. "So the companies are pretty clearly trying to make a global kit, and we are getting the benefit of that," he said.
Hopes for NGS, but challenges remain
Other advances remain further off. NGS, in particular, has drawn significant interest from forensics researchers, but, Myers said, while "people are certainly putting a lot of effort toward this," he has yet to see evidence that it's ready for use in actual casework.
NGS's potential, however, "is astonishing," Kobilinsky said, noting that the technique will allow investigators to look at much larger numbers of STRs as well as other genomic features like single nucleotide polymorphisms, which will enable both higher accuracy in matching samples and in determining relationships between samples. It could also improve and streamline mitochondrial DNA analysis, which is currently done using Sanger sequencing.
NGS could prove particularly useful for analysis of degraded or contaminated samples, suggested Cydne Holt, senior market manager, market development, forensics at Illumina, which, along with Life Tech, is one of the main vendors pursuing the application of the method to forensics.
Often in the case of degraded samples, only small fragments of DNA are available, leaving investigators with only enough material to generate a partial profile using conventional STR analysis, Holt said. This weakens the statistical power of the technique, leading, in theory, to less conclusive identifications.
"There is demand for multiplexes of markers that are less than 150 bp to address highly degraded or PCR-inhibited casework samples," she said, adding that "this demand is increasing as laboratories are requested to interrogate more and more challenging samples."
In the case of mixed, or contaminated samples, NGS could help distinguish between DNA from different sources such as victim and perpetrator by analyzing a much higher number of genomic features than just the 15 STRs tested under the current standards.
Given NGS's ability to generate phenotypic data, it could also prove useful as an investigative technique or for developing evidence to corroborate eye witness testimony, Holt said.
"SNPs and repetitive elements can be genotyped at the same time to assist investigators [in] uncovering how a perpetrator may look or what they may have recently been doing," she said.
Butler agreed that, in theory, NGS could allow investigators to "get a lot of additional data from their samples" including information such as eye and hair color. However, he noted, it remains to be seen "how well that will work in practical terms." He also raised the possibility that use of the technique for such a purpose could run afoul of privacy concerns.
As reported in June by In Sequence, researchers from the University of North Texas Health Science Center provided a glimpse into the future potential and remaining challenges for NGS in forensics in a study published in the International Journal for Legal Medicine in which they compared the performance of Life Tech's Ion Torrent PGM instrument and Illumina's Genome Analyzer for analyzing SNPs in a forensics setting.
The team, led by UNT professor Bruce Budowle, an expert in the use of NGS for forensics and one of the original developers of the FBI's CODIS database, used the two platforms to investigate 95 SNPs common to both Life Tech's Ion AmpliSeq Human Identification SNP panel and a SNP panel run on the Illumina GA. They also tested the Life Tech panel on samples ranging from as much as 10 nanograms of DNA to as little 100 picograms in order to test how it might perform with the sort of small sample sizes often found at crime scenes.
The two platforms agreed on 94 out of the 95 SNPs, with the one discordant result stemming from a PGM sequencing error. With regard to sample size, at 10 nanograms the PGM identified all the SNPs in the AmpliSeq panel, but at 100 picograms – a not uncommon sample size in forensic analysis – it missed an average of 1.6 SNPs per sample, and in some samples missed as many as six.
Speaking to In Sequence upon release of the paper, author Jianye Ge, an assistant professor at UNT's Institute of Applied Genetics noted that while the technology was "very promising," it would "need to be developed more" before it would be useful in actual criminal investigations.
Beyond questions about its accuracy and sample requirements, NGS also remains more costly and time consuming that traditional STR analysis, Butler said. "Until those things are improved upon, [the field] probably won't be switching to [NGS]."
"There is a really big difference between research activity and putting something in casework," Prinz said. "And a lot of research activity never gets anywhere."
In addition to NGS, Prinz cited DNA microarrays as another busy area of forensics research, with firms like DNA Link and Identitas working on microarray-based systems for identifying individuals using SNP analysis. On the mass spectrometry front, Abbott's PLEX-ID system, which it acquired through its purchase of Ibis Biosciences in 2009, combines PCR with electrospray ionization mass spec to identify DNA amplicons. The instrument's future as a forensics platform is uncertain, however, as Abbott announced last year that it was discontinuing the original version of the system to concentrate on a next-generation instrument aimed at the clinical diagnostics market.
Proteomics has also made some inroads into the forensics space, though far less than have genomic techniques. The discipline will likely prove most useful for distinguishing between different bodily fluids, such as menstrual blood and venous blood, Prinz said. It is also being explored as a method for detecting protein polymorphisms in hair samples, where DNA is present in low amounts. In a sense, Prinz noted, forensics research into proteomics harkens back to the field's early days, as protein-based blood typing has now been used in criminal investigations for almost 100 years.
"There's a lot of exciting stuff coming down the pike, and we're looking at these things in our research department," Myers of the California Department of Justice said. At the same, though, he noted, "a lot of 'next best things' come out that are highly touted ... and we have been around long enough that we've seen a lot of these 'next best things' come and go."
"I just don't know how practical it will be do these types of investigations due to the nature of the samples we look at, the questions about how accurate the results are ... it's a lot to work out," Sims said. In forensics "you really want to make sure all your ducks are in a row, because this is a very serious challenge. To put someone on death row based on a scientific test, it has to be good science."
"On the other hand," he acknowledged, "you of course never want to bet against technological advancement."
For the time being, though, the criminal forensics landscape, at least in the US, appears relatively settled, Myers said, with the techniques currently in use fairly well established and newer approaches not yet "making a big enough dent that they are getting much court attention."
And while that might sound dull to enthusiasts of the cutting edge, the situation, he noted, does have its advantages – not being stuck for days on end in evidence admissibility hearings, for instance.
"It's been a pleasant time, going through the last 10 years or so without having a huge number of challenges, just being able to introduce evidence in court," Myers said. In fact, he hinted, his younger colleagues who've come up during this lull don't know how good they've had it.
"Most of the newer criminologists are used to [giving] 15 minutes of testimony instead of 24 days," he said.