Forensic Phylogenetic Analysis: Under the Microscope
Experts search for best way to present phylogenetic findings to a jury
In 1994, in Sweden, phylogenetic analysis was first used in court to help obtain a conviction in a case of HIV transmission. Since then, phylogenetic analysis—sometimes given the misnomer “HIV fingerprinting”—has been used repeatedly in world courts to try cases of deliberate HIV transmission—cases where a person is charged with deliberately infecting another with the HIV virus, either through unprotected sex, dirty needles, or blood exchange.
As with most scientific evidence, the best way to present phylogenetic findings to a jury is still evolving. Experts themselves don’t always agree.
Molecular phylogenetics is a means by which the genealogical pedigree of micro-organisms can be reconstructed from information found mostly in their DNA. The most common forensic use is for tracing or dating transmission of viruses. If organisms can be shown to share a common ancestor, this information may be forensically useful, especially in cases involving intentional HIV transmission.
“Many viruses have high rates of evolution, so there are enough genetic changes in the viruses that they can be followed through time and across transmission events,” said David Hillis, a professor of Integrative Biology, University of Texas.
Forensic phylogenetic analysis works like this. Blood samples are collected from individuals in question, as well as from a control group. The samples are coded, blinding the study to investigators. Viruses are extracted from the blood and DNA and RNA sequences are obtained from the viruses. The analysis, which takes a few days, is usually performed by an expert in phylogenetic analysis and computational biology. That expert, who only knows which samples are the controls, determines if one of the test samples can be identified as being consistent as the transmission source. This is done by determining if a recipient shows a phylogenetic subset of the sequence variation present in the source. Once a determination is made, identities are decoded and the results found to support or refute charges.
In a 2010 paper (Proc Natl Acad Sci 2010 Dec 14), Hillis said phylogenetic methods are ideally suited for determining the HIV pattern of descent in cases of suspected transmissions, and that phylogenetic analysis can also provide evidence about the direction of transmission. Direction of transmission is supported by a phenomenon called paraphyletic relationship—where a subset of source viral sequences is more closely related to all recipient sequences than to other source sequences.
"Strongly supported paraphyly can provide evidence to infer direction of transmission between pairs of epidemiologically related individuals," Hillis said. "However, a lack of paraphyly cannot be used to refute a possible transmission route."
Hillis, and others, caution, however, there is much potential for abuse in current phylogenetic analysis methods. It is inappropriate, for example, to use phylogenetic analysis to survey databases to look for closest matches, and then use that information to make transmission accusations. Phylogenetic analysis is not like a DNA fingerprint, in that there is no expectation of an exact match between transmission pairs.
“Phylogenetic analysis is appropriate only when there is an explicit a priori hypothesis to test,” Hillis said. “It should not be used to search a database for closest matches, as is done with DNA fingerprinting.”
Shortly after the Hillis paper appeared, European medical researchers made the claim that evidence obtained through phylogenetic analysis can never be conclusive, in that the conclusions do not preclude reasonable doubt.
In an editorial in the February 2011 issue of The Lancet, six European researchers agreed that phylogenetic analysis is a powerful technique that can, if properly used, provide valuable circumstantial evidence in HIV transmission cases. However, forensic scientists should be aware of the limitations of this analysis and emphasize that courts must use ancillary evidence to achieve convictions.
Anne-Mieke Vandamme, of the Laboratory for Clinical and Epidemiological Virology, Rega Institute for Medical Research at the Katholieke Universiteit, Leuven, Belgium, told Forensic that phylogenetic evidence is never conclusive.
“Phylogenetic evidence is at best circumstantial and it is at its strongest when the evidence is against the case,” she said.
Vandamme maintains that when phylogenetics finds a monophyletic cluster joining victim and suspect, this evidence cannot prove transmission beyond reasonable doubt. Proper identification of the transmission source would require that a phylogenetic tree can flawlessly reconstruct a true epidemic history and that strains from all patients ever infected with HIV are available as controls—both unrealistic assumptions.
Thus, since the full transmission tree cannot be known, no proof of direct transmission can be claimed, even if the hypothesis cannot be rejected. Conversely, separate clustering with unlinked individuals can disprove direct transmission.
Science in Court
Hillis responded that The Lancet editorial authors criticized “HIV fingerprinting,” which is not the same as phylogenetic analysis. HIV fingerprinting attempts to identify a specific HIV sequence, not to reconstruct a transmission history.
“They also greatly over-stated the case,” he said. “Although some points made are valid, they are also standard operating procedure in phylogenetic transmission investigations.”
For example, it is important to have a clear a priori hypothesis to test, and it is important to blind the identities of samples during the analysis. Clearly, a case cannot rely merely on phylogenetic analysis, Hillis said. There has to be clear epidemiological evidence and a criminal investigation and forensic standards of investigation must be maintained.
Hillis does not dispute that phylogenetic analysis represents circumstantial evidence. “All scientific inferences from data, of any kind, represent circumstantial evidence, by definition. That’s the way the analyses are presented in court,” he said.
Where Hillis disagrees is that there is any assumption that the phylogenetic tree is inferred without any error. Rather the error is estimated and reported, the same as any other analysis of scientific data, Hillis said.
He also disagrees that everyone who has ever been infected with HIV must be sampled. Rather, conclusions can be supported or rejected on samplings actually performed.
“Given these points, there is no doubt that a phylogenetic analysis of multiple strains from a victim and a suspect, as we conducted in the PNAS paper, can indeed support or reject particular hypotheses of transmission between suspected transmission pairs,” he said. “The standards of such analyses have been tested and supported in courts of law.”
Hillis said no genetic test can ever provide 100% proof of a hypothesis. Science does not work that way.
“But, the tests can reject or support particular hypotheses to a high degree of significance, just as for other types of genetic testing,” he said.
Vandamme, who has testified in over 20 HIV cases, worries that juries may not understand the scientific meaning of "support" of a transmission hypothesis “to a high degree of significance,” in the context of a priori hypothesis testing.
“I would be more careful to state that the hypothesis of transmission can be rejected or not with a high degree of significance using phylogenetic analysis techniques,” she said. Vandamme believes, since phylogenetic analysis alone cannot prove HIV transmission, a statement that phylogenetic analysis supports a transmission hypothesis will convey the impression that phylogenetic data are proving the transmission, even if it is specifically stated that phylogenetic analysis alone cannot prove transmission.
“This is a common misinterpretation,” she said. “Rather than stating that the phylogenetic analysis is supporting a transmission hypothesis, it is entirely correct and more prudent to state that the transmission hypothesis cannot be rejected.” There is much more confidence in rejecting the transmission hypothesis, she said, when the viruses are found not to be related than accepting the hypothesis when they are very closely related, even in the context of a priori hypothesis.
Hillis agrees it is important to be clear about what the analyses can and cannot support.
“That point,” he said, “is not unique to phylogenetic analysis—it is true of all scientific analyses.”
Douglas Page writes about forensic science and medicine from Pine Mountain, California. He can be reached at firstname.lastname@example.org.