HKU Bulletin October 2008 (Vol. 10 No. 1)

fluid. Then, the DNA of both samples is extracted, studied, and compared. If the DNA matches, then the suspect can be placed at the crime scene. Since the late 1980’s, DNA results have been widely used as evidence in the courtroom - perhaps most sensationally during the 1995 O.J. Simpson trial. At HKU, Personal Professor in the Department of Statistics and Actuarial Science, Dr Tony Fung Wing Kam has spent much of the last decade developing methodologies to help government forensic scientists interpret DNA samples. Recent winner of the prestigious Croucher Senior Research Fellowship award, Fung said that statistical methodology plays a key role in ensuring that DNA evidence is analysed, interpreted and presented correctly. And with the recent advances in computer technology, this methodology is more complex than ever before. He came to the area of research quite by accident. “In 1993,” he said, “the department head, at the time, asked me to teach a course on Statistical Applications and one application was in law, but I had no idea how to teach it, because I was not doing law. “We discovered that DNA profiling is one statistical aspect of law, so to find out more I went to a conference in Arizona. Although I participated, I had little idea of what they were arguing about. I had no idea, at the time, exactly what DNA profiling was, so I asked a number of simple and naïve questions. “After I came back I contacted the Hong Kong Government Laboratory because they were also working on this area from a forensic, or legal, point of view. They were also very aware of the chemical or biological side of DNA profiling. I consulted them frequently and began to learn more about this area and to work on some simple problems,” he said. At the time, in 1994, a newborn baby was stolen from the Caritas Hospital. A few days later a newborn was found abandoned in Yau Ma Tei. “But we couldn’t identify the baby,” said Fung. “All newborns look similar, and also all Chinese babies have black hair, so there was no way of proving its identity. “So the government’s scientists needed to do the DNA test and asked me if I could do the calculations.” The situation he said was somewhat different to regular paternity tests, which usually involve the mother seeking to prove the baby’s paternity. “But here we needed to confirm that the couple were both parents to this baby. The government did not know how to do the calculation, so I provided the formula and then did the statistical assessment.” 17 From there grew his fascination with both the practical and theoretical aspects of DNA profiling. Since then his work has been published in several journals and he and his PhD graduate have recently published a book, Statistical DNA Forensics: Theory, Methods and Computation . It explains much of the recent developments in statistical methodology for DNA analysis. In recent years he has concentrated on the complex problem of DNA mixture. And he explained: “If you find a blood stain on the carpet at a crime scene, we might assume that it was left by the perpetrator, and not the victim. If we compare the DNA of the suspect with the blood sample and they match then that’s simple. But by law, we have to provide figures, which means devising a theory, or method and then a computer programme to do the calculations. “However, this sort of case deals with single person DNA and is therefore relatively straight-forward. Mixture, in contrast, involves the DNA of more than one person. This makes it more difficult to analyse because you can’t separate the blood. For example, during the OJ Simpson case, in which two people were murdered, blood stains were found from two people, but it was impossible to separate them. This mixed DNA also occurs in multiple rape cases where there may be a mix of several different DNA’s. “You have to look at several chromosome loci to narrow down the likelihood of the DNA belonging to one particular person. You can imagine how difficult this is to do when two, or more, people are involved.” The difficulty increases when the samples are from individuals of a different racial background. “In the OJ Simpson case there was a blood mixture, but at that time no method had been developed to distinguish between them so they could only assume that both samples were from men of the same race. So you can say that the figures they provided at the time were probably wrong.” That spurred on Fung to develop his own statistical method, or formula, to assess different possibilities. “After all the fuss of the OJ Simpson trial I recognized there was a need for a new method,” he said. He is now working on developing a theory for calculating DNA mixture from the government’s offender’s database. “This is my main area of research,” he said, “and I will also be collaborating with the Hong Kong Government Laboratory by providing them with scientific figures, methods and computer programmes to do the analysis.” RESEARCH DNA Forensics and the Law How science fights crime F or most of the last century fingerprinting was considered one of the most accurate ways to place a suspect at the scene of a crime. The revolutionary technique succeeded in putting many criminals behind bars. Today, however, a more accurate, and equally revolutionary, method is frequently used: DNA evidence – often considered the silver bullet in the trial lawyer’s armoury. The process is simple: two samples are taken, one from the suspect and one from the scene of the crime. The samples may be skin tissue, hair, blood, semen or vaginal 16