DNA Science Workshop for VCE Biology Teachers
But you might be surprised to learn that the case is also being intently watched by the scientific community. That's because, besides O.J., the science of DNA fingerprinting is also on trial.
Preliminary DNA tests show that Simpson's blood matches that found at the crime scene. Simpson's defense attorney's are aiming to keep that evidence out of the courtroom. And they may have a reasonable argument for doing so. California's courts admit scientific evidence only if it is derived from methods that are supported by a scientific consensus -- the so-called Frye Kelly standards. In the pre-trail admissability hearings Simpson's defense will argue that there is in fact not a scientific consensus about the use of DNA fingerprinting in the courtroom.
There's no denying that controversies have abounded over the last few years over the forensic use of DNA fingerprinting. A scan through the journals Nature and Science shows a smouldering record of the skirmishes between scientists.
I'm going to try and explain to you what these skirmishes are about. But for starters, I'd best explain what a DNA fingerprint is.
Let's start with a traditional fingerprint. No two individuals have an identical fingerprint, not even identical twins. So if a suspect's fingerprints match those found at the scene of a crime, its certain that he or she had something to do with it. But fingerprints are not always retrievable from a crime scene, so forensic science has sought to take advantage of another unique characteristic of a human being, the genetic blueprint as carried by the DNA of each cell.
An individuals genetic blueprint is unique. But it is far beyond the bounds of feasibility that one could get a readout of the entire DNA sequence from a forensic sample.
So DNA researchers have worked out the next best thing. A DNA fingerprint samples an individual's DNA sequence. If we think of a person's genetic blueprint as a text, then the DNA fingerprint is like a sampling of the first word that might occur on say pages 16, 48, 123 and 200. No two texts are likely to have the same word at these four positions. The combination of the four words would be distinctive for the individual text. A persons genetic blueprint can be sampled in the same way. At certain locations of the genetic blueprint, the DNA sequence is known to vary greatly from person to person. These sites are referred to as VNTR markers (for variable number tandem repeats). I'm just going to refer to them as markers. A DNA fingerprint is generated by sampling the person's DNA at four to six of these markers to build up a distinctive profile. But is the DNA fingerprint unique? In most cases it is not. To calculate the rarity of the fingerprint, one has to know the frequency with which each type of marker occurs in the population.
I hate to do this to you but I'm going to switch metaphors. Get out your imaginary card pack and we're both going to do the same thing. Sort the pack into four piles according to the suites and make sure each pile is well shuffled. Now pick a card from each pile. I've drawn a Queen of hearts, a King of spades, a two of clubs and a six of diamonds. That's the card fingerprint I've generated The chance of your matching it is most unlikely, 1/ 13 to the power of 4 or one in 28, 561. Like the 13 different members of each suite, each VNTR marker chosen for DNA fingerprinting occurs at a particular low frequency in the population. Multiplying out the frequencies for four different markers typically gives a DNA fingerprint with a frequency of between one in 100,000 to one in a million.
So let's talk now about what happens in a real case. There's been a rape. The forensic team have recovered semen from the victim and carried out DNA fingerprinting. It's a highly distictive DNA fingerprint; only one in a 100,000 people would be able to match it. Police round up the suspects; most are cleared outright because their DNA fingerprints don't match. But one of the suspect's has a matching fingerprint.
At his trial the forensic expert tells the court that the defendant's DNA fingerprint matches the one found in the semen and that only 0.001% of the population, or one in a 100,000 people would be able to match it. That's a highly distinctive test, it would exclude 99.999% of the population.
That's certainly bad news for the suspect. But it does not provide absolute proof of guilt. The court uses that information along with all the other circumstantial evidence presented at the trail to arrive, if it does, at a verdict of guilty beyond reasonable doubt. One of the things the court has to decide is how many men with that DNA fingerprint there actually are who could have committed the rape. If there is no grounds for narrowing down the population of suspects, the court might have to consider every eligible rapist in Australia, some four and a half million men. As one in a 100,000 would have that DNA fingerprint, they would have to consider that 45 individulas in Australia could have committed the crime Of course in most cases, there are grounds for narrowing down the population of suspects. So if the rapist in our example was known to be a local man in a town of 3000, the odds of one in 100,000, would overwhelmingly incriminate him.
In a nutshell that's how DNA fingerprinting is carried out and used in the courtroom. Now what's all the controversy about.
There are two issues. The first is essentially about "quality control" in DNA testing
When DNA fingerprinting was first introduced into the courts in 1988, it was seen as infallible. The technology was straightforward; it was the tried and tested workhorse of gene cloning. And the theoretical premise was self-evident; everybody's DNA sequence was unique; it was just a matter of sampling enough of the distinctive markers.
But the problem came in transferring the technique from research lab to commercial laboratory. The standards required in the two situations are different. A research scientist repeats her experiment many times before drawing any inference. In the forensic setting an inference often has to be drawn on the basis of one experiment.
Most of the problems attending the birth of commercial DNA testing were exposed by MI T geneticist Eric Lander in a 1989 Nature article. No doubt the paper caused a collective cringe in the scientific community as all the vulnerabilities of science's brightest baby were luridly exposed. Lander had served as an advisor for the defence in several cases, the most famous of which was the Castro case. In that case, DNA fingerprinting carried out by the company Lifecodes, incriminated the suspect. But Lander as a witness for the defence found that the technician had used a number of subjective judgements about whether the two DNA fingerprints matched. Those judgements proved indefensible under Lander's rigorous scrutiny. The judge finally ruled the DNA evidence inadmissible.
Lander highlighted the room for subjective error in forensic DNA fingerprinting and urged the scientific community to establish clear guidelines and standards to ensure that the procedures were reliably carried out. Many scientists, Lander included, consider that these procedural problems have been largely fixed. The major laboratories now work according to standardized procedures and subscribe to voluntary proficiency testing and the guidelines are set to get even stricter.
The continuing bone of contention lies in calculating the rarity of the final DNA fingerprint.
In DNA fingerprinting its the final statistic delivered to the courtroom that carries the punch. When two fingerprints match with a statistical likelihood of one in a million, that carries a lot of convicting power.
So its not surprising that the means of obtaining that statistic have been open to intense scrutiny both by scientists and defence lawyers.
Like I said before, to calculate the rarity of a person's DNA fingerprint, you multiply the frequencies of the individual markers used, just like we did with the pack of cards.
However just like with the cards, some assumptions are made.. When we generated the card fingerprint, I was certain that the probabiltiy of each draw from my pack (1/13) was the same as the probability of a draw from your pack. We were also certain that the card we drew from each pile had no influence on our other draws, i.e. the draws were independent.
With DNA markers the same sort of thing is assumed -- that the frequency of a particular type of DNA marker is pretty much the same whatever population you sample and, that the markers are independent of each other, that is if one individual carries marker X, he is no more or less likely to carry marker Y.
But these assumptions may not be true of most Western populations. Harvard geneticists Daniel Hartl and Richard Lewontin raised this concern a few years ago. They pointed out that the U.S. population is largely composed of migrants, most having come since the early 1900's. And that the racial group considered as Caucasian is actually made up of ethnic groups that have not really mixed their genes -- Jews, Irishmen, Italians, Poles, Swedes and so on. Lewontin showed that within these groups, there can be significant differences in the frequency of the same gene. In fact, after tracking seventeen genes , Lewontin concluded there was more genetic variation between Caucasians, than between Caucasians, Africans and Asians.
That raises some theoretical problems. The rarity of a DNA fingerprint found at a crime scene is calculated using the frequency of markers for the general population data base. But lets say that one of the suspect's with a matching DNA fingerprint belongs to an ethnic group where that DNA fingerprint is more common. The court might be told that the DNA fingerprint would be found with a frequency of one in a 100,000 in the general population, yet in that person's ethnic group it might occur with a frequency of one in a 1000. Those are certainly far less convincing odds to convict someone with.
This is the issue on which battlelines have been drawn. Many notable geneticists and the FBI have spoken out declaring that even though there are ethnic subgroups, there do not appear to be significant differences in the frequencies of the DNA markers used for fingerprinting, and that certainly for the purposes of the courtroom, current procedures are adequate. Others notables remain concerned and believe that the data bases for the different ethnic groups need to be known, so that the rarity of a DNA fingerprint can be determined in say an Aboriginal, Vietnamese or Italian comunity. They also believe that until such data bases are available, a ceiling needs to be set on the statistical power of the final DNA fingerprint. Individual markers should be used at a frequency of no lower than 10%, or at the highest frequency found in any subgroup of the population.
There is however one point on which all camps agree. As the number of individual markers tested increases, so does the statistical power of the DNA fingerprint. By the time eight markers are tested, even using the conservative ceiling principle, the rarity of the fingerprint gets close to one in a billion, and the blows away any statistical issues. That's precisely what O.J. Simpson's prosecution have done. They have sent his DNA to two different laboratories, with some eight different markers being tested.
That blows away any claims of laboratory error and the statistical issues in one go.