Caught with suspicious genes

A DNA database for criminal records is about to be launched by the Home Office. Wilson Wall weighs up the pros and cons

Next Monday, the world's first DNA database for criminal records will open for business at the Home Office's Forensic Science Laboratory in Birmingham. Over the next year, it is expected to receive 135,000 samples from people suspected of burglary, serious assault and sexual crimes.

By the end of the century, however, up to 5 million offenders could be logged on the computer files, as chief constables plan to expand the service to cover all recordable offences - those punishable by imprisonment.

Police forces are embarking rapidly on this new genetic technology, even though less than a week ago they were forced, embarrassingly, to abandon a computer database for analysis of traditional fingerprints upon which millions of pounds of taxpayers' money has already been spent.

Every cell in the human body, with the exception of red blood cells, contains DNA. Any cell can therefore be used for DNA analysis. Just as we are all different, so our DNA is different (with, of course, the obvious exception of identical twins).

The earliest applications of DNA analysis used a relatively imprecise technique which had a number of drawbacks. It required undamaged DNA, whereas bacterial and fungal decay can damage DNA in forensic samples. There were problems with laboratory procedures. This was more commonplace in the United States, where it has been admitted that sloppy work and contamination blighted presentation of data at court.

But the issue that caused consternation was the manner in which the final results were presented. Calculations were made of the probability that a particular result might have appeared by chance and the accumulated statistical values were ridiculous. One American laboratory reported the possibility of a random match as one in 738 trillion! What had eluded the person who calculated this number was that it was so much larger than the total population of the planet it implied the profile was unique. Yet the biology underlying the technology tells us a different story - that no profile is unique.

One of the live issues in the Los Angeles trial of OJ Simpson is whether DNA evidence is admissible, on the grounds that a lay jury could have difficulty understanding the probabilities. The history of DNA profiling suggests that even scientists have had difficulties with the numbers.

By the end of 1994, a major change had taken place in the method of DNA analysis. It solves some of the earlier problems, although it raises new ones of its own.

Two discoveries have completely altered the way DNA is analysed. The first was the discovery of "short tandem repeats" (STR). These are repeated sequences of DNA which do not actually get translated into proteins but lie within and between active genes. The repeat motif is usually three or four "letters" of the genetic code and typically may be repeated about 20 times. What is particularly useful about STR analysis is that different individuals have a different number of repeats.

The second major breakthrough came about when the American Kary Mullis devised the Nobel prize-winning invention known as the polymerase chain reaction (PCR). This is a test tube-based method for taking single molecules of DNA and "photocopying" them billions of times while maintaining extreme accuracy of replication. Having amplified the volume of DNA from a particular sample, it is then possible to analyse it to measure the sizes of the STRs it contains.

Forensic science was quick to grasp the potential of a technique that could be used on samples comprising only a few cells. STR analysis can be made on all kinds of human material, from the tiniest scraps of dandruff to old and damaged samples of blood and sperm. DNA will survive long after death, even when the body is waterlogged or buried, which makes it ideal for use in forensic applications.

There are other advantages. Suspects need not give blood samples, a process that requires either their consent or a court order. A simple mouthwash will yield enough cells for the STR test.

However, STR profiling is not infallible. Profiles can be identical, which means the suspect might be innocent but have the same profile as the criminal. Therefore, STR analysis should not be used on its own to send someone to jail. What it can do is rule out a suspect as the criminal - when two samples do not match, a result known as an exclusion.

Another potential stumbling block is the statistical interpretation of the findings. A calculated probability of, say, one in 20,000 is the probability of finding a particular STR profile by chance. This is not the probability of either guilt or innocence, a subtlety that sometimes manages to elude lawyers and juries.

Research shows that different lengths of STR are common in some ethnic groups and rare in others. This means that sensible calculations can only be made by comparing the individual's STR profile with a comparison population identical or close to their ethnic group. In isolated cases, a comparison population may not be available for the suspect being tested. This could invalidate the whole process, since a rare frequency in one group may be extremely common in another.

While scientists argue over the statistical interpretation of results, others are deeply uneasy about the ethics of a DNA database. At the moment, data from samples can only be kept if a suspect is found guilty. These samples could save a great deal of police and court time if the suspect re-offends. But some people are calling for the system to be extended. They want data to be collected and kept even from citizens without a criminal record. It is this that causes such disquiet.

Ethics apart, there is a good economic reason for not proceeding with a national DNA database so precipitously. In 1986, the "old" techniques of DNA profiling were used for the first time in a British court to link two sexual assaults and murders in Leicestershire. Now courts use the "new" STR analysis. The two systems do not yield the same information, so trying to compare results from old DNA profiles with STR profiles is the forensic equivalent of comparing apples with pears.

As the Home Office Forensic Science Service sets up its database using STR technology, it will be tied into these systems for a very long time. However, in a few years, STR analysis may be superseded by other, even more sophisticated techniques, invalidating any earlier databases.

Dr Wilson Wall acts as scientific consultant to the defence on the use of DNA evidence in court.