DNA testing is not infallible

SINGAPORE, Sept 6 — Last week, Malaysian opposition leader Datuk Seri Anwar Ibrahim was sworn into Parliament. He then led a walkout over a Bill that would compel suspects to surrender samples of their DNA.

That fight seems almost quixotic given that the technology is so sensitive now that DNA in trace samples with just a few cells will suffice. There is enough of your DNA on the newspaper in your hands now for a crime lab to profile it. Assuming there is no police misconduct, the Anwaristas ought to be fighting instead for a better law, one that ensures only good DNA evidence is admissible in court.

Contrary to popular opinion, DNA evidence is not unassailable. There is both good and bad DNA evidence.

The DNA Profiling Lab at the Health Sciences Authority here says its error rate is only one in trillions. But last month, police in Victoria, Australia, admitted to an erroneous DNA test that had secured a rape-and-murder conviction. It then reopened 7,000 convictions over the past 20 years based on DNA evidence.

Like any lab procedure, DNA testing can produce false negatives. That is, a match that exists in reality is missed and a guilty person is exonerated. Worse yet, there can also be false positives, where an innocent person is wrongly convicted.

Because of the high statistical standards used in DNA testing to establish similarity between samples, the risk of false positives is infinitesimally small. But statistically, this means that the likelihood of false negatives is, by comparison, relatively higher.

How do errors occur? Accurate computerised equipment is used for the testing — leading to colourful graph printouts — but these graphs must be interpreted. Judgment calls are necessary as to which peaks in the graphs indicate actual genes and which are artefactual.

But these colourful printouts can have bleed-throughs. A peak labelled with blue dye, say, might be mistaken for a yellow or green one, thus creating false peaks (genes) at the yellow or green.

While true peaks are narrow spikes, false ones can be blobs. Because of this ambiguity — how much wider are blobs? — crime lab analysts must interpret the printouts. So how they do so should be closely scrutinised.

It is also important to rule out sample degradation, contamination during collection, cross-contamination by other samples in the lab, or simple handling errors like mix-ups or mislabelling.

Even without such mistakes, DNA evidence can still be misleading because much of the technical details behind commercial forensic DNA test kits which crime labs invariably use cannot be scrutinised. The kits come complete with all the reagents required and instructions on how to run the tests. However, manufacturers don't divulge any other technical details about the kits because they say these are trade secrets. They have fought off subpoenas in the US to reveal such information.

Kit makers are assumed to carry out tests on their own kits to verify their reliability. But this data is not divulged because there is no law that makes it a requirement.

The other set of information that kit makers don't like to reveal is the exact composition of one of the reagents used in their kits called “primer sequences”. Primers are used in the step called polymerase chain reaction in which DNA is “Xeroxed” millions of times. (That is why only trace amounts of DNA are now enough for profiling.)

But if the make-up of these primer sequences cannot be scrutinised, then one cannot be sure if the xeroxing they enable is accurate. Imagine xeroxing a million copies of a wrong page of text submitted as evidence in court. The evidence would still be wrong. Thus knowing the primer sequences that kits use is crucial if we are to be sure that a suspect's sample is correctly xeroxed.

All in all, the justice system must scrutinise the nitty gritty of DNA reports more critically, including even kit reliability. And legislatures must write DNA laws to compel third-party kit makers to release such information if they want to continue to do business in the country. DNA may well be the gold standard in forensics but the dross must be skimmed off first. — The Straits Times Singapore