Suppose you are a juror faced with a man accused of a sexual offence, and you are told his Y chromosome DNA profile matches a sample obtained from an incriminating location at the crime scene. However, there is little else in the way of solid evidence to implicate him. Is that enough evidence to vote ‘guilty’?
Before making such an important decision you’d expect the court to provide some guidance as to how strong the evidence is. After all, Y chromosome profiles have been used internationally in criminal courts for about 20 years, and so there is considerable accumulated experience in legal systems.
Unfortunately, in practice, courts give jurors little help to assess the strength of this evidence. To help address this, Associate Professor Mikkel Andersen from Aalborg University in Denmark and I have developed a new approach, published in PLOS Genetics, to explaining Y profile evidence that is scientifically robust, fair and easy to understand.
How Y chromosome evidence is currently used
Although Y profiles are much less discriminating than standard DNA profiles, which use information from many chromosomes, they retain a niche role in sexual offence cases where a trace amount of male-origin DNA is ‘swamped’ by female-origin DNA. Looking only at the male-specific Y chromosome makes the female-origin DNA invisible, in effect.
Perhaps surprisingly, there is no widely accepted way to convey weight-of-evidence for Y profile matches. That’s not for want of trying, as there has been an extensive academic literature on the topic over two decades. But even the most authoritative US and European recommendations for Y profile evidence, while strong on technical details of the sample handling and profiling protocols, are vague on the crucial issue of evidential weight.
Most often courts are simply told the frequency of the Y profile in a database, which in many cases is zero. But the databases are small relative to the population, and they are ad-hoc collections rather than scientific random samples, so this information is not very helpful.
Some convert the database count to a population match probability, analogous to that reported for a standard DNA profile, where for example a prosecutor might say, “there is only a one in one million chance that a random man would have a matching profile”. But specifying the relevant population is problematic and, for reasons given below, Y profile match probabilities are much more sensitive to this choice than standard DNA profiles.
Advising defences, I’ve often found it easy and a moral duty to criticise prosecutions for such potentially misleading approaches. But as an academic making occasional court appearances, I haven’t had the impact to change practice. And I’ve also not had a detailed solution to offer in place of current poor practice – until now.
The problem with Y chromosome evidence
The fundamental problem that neither courts nor academic papers have been able to satisfactorily take into account is that a man shares his Y profile with his father and his father’s father and so on, going back possibly many generations. He will also share it with most of the living male descendants of these men – perhaps even thousands of them for older Y-profiling systems.
Newer systems are more discriminating (by exploiting the high mutation rates of some Y-chromosome markers) and so the number of matching male relatives is now rarely more than 100. But that can actually make the problem of fair evaluation of the evidence even harder.
Those 10 or 50 or 100 matching men are not random members of the population but an extended patrilineal ‘clan’ almost certain to be missed by small and unscientific databases, and likely to resemble the defendant in some respects (location of residence, appearance, ethnicity) that may enhance the possibility of a false accusation. Some of the matching clan members may be separated from the defendant by more than ten father-son steps, more distant than the relatives you’d see at even a large family reunion, but for a random pair of men the probability to be so closely related is much less than 1 in 1,000.
The chance of completely unrelated men having matching Y profiles – which previous methods have focussed on - is actually so small with modern profiling systems as to be negligible.
Associate Professor Andersen adapted his existing software so we can now tell courts about the likely number of matching males in the population, and the possible consequences of their relatedness.
Based on simulations of population history, the software investigates the numbers of matching men under different scenarios. We can also adjust the result to take account of database information – but this effect is often surprisingly small.
The court must then decide if it has enough other evidence to identify the suspect as the source of the crime-scene Y profile, rather than one of his matching (distant) relatives.
Our approach is radically simple: what can be easier to understand than a statement such as “there are probably no more than 40 other men with the same Y profile”? The complexity lies in the underlying population genetics models, which require assumptions about population size, growth rates, variance in reproductive success and the mutation processes at the relevant Y chromosome markers.
But the effect of such parameter choices is often modest, and it is possible to report an upper bound that applies under a range of reasonable assumptions.
We believe that our new approach is scientifically more robust than previous approaches, and easier for courts to understand and to probe.
Banner image: iStock