Round up all the Fortescues! DNA crime scene surname matching

DNA could be used by police, today's public prints tell us, to 'predict the name of suspects', according to a new study from the University of Leicester Department of Genetics. And indeed it could, but only up to a point, and the way the police would do it involves different, slightly worrying, ways of looking at the data, rather than any new gosh-wow breakthrough in DNA analysis.

The study, published in Current Biology (abstract here) found that in a sampler of 150 pairs of men with the same surname, in 25 per cent of cases the pair had matching Y chromosomes. When the common surnames such as Smith and Jones were excluded, this increased to 50 per cent, and when you have large numbers of DNA samples on record (the UK database is now over 3m and climbing), the potential here starts to become clear.

Familial DNA matching has already been used by police in the UK where the DNA found at the scene of crime was of an individual who wasn't on the database, but who was related to someone who was. But if police were to begin to associate their existing samples by surname or groups of surname, then a similar approach could be employed on a far wider basis. So, round up all the Fortescues?

Kind of. If for the past 800 years or thereabouts everybody had behaved themselves in bed (or haystacks or whatever), and if naming systems had been completely orderly, as the UK database grew it would start to spawn detailed and precise family trees, and the surnames of mysterious offenders could be predicted with a very high rate of accuracy.

Real life isn't quite like that, but it will still be possible in some cases to produce a suspect's surname from the DNA sample - provided the police database jockeys get their sums right. Police will then be able to prioritise elimination sampling*, so instead of (as is current practice) requesting that, say, the entire adult male population in a specific district provide samples to 'eliminate them from the inquiry', they can just ask those adult males called, say, Hayman first. And widen the net if this doesn't throw up anything? To the whole local population, or to Haymans further afield?

Discovering you're the owner of a suspect surname will clearly be a tad unnerving, and the approach is obviously dangerous in that, if the database jockeys aren't doing their jobs properly and the technique is used as a blunt instrument, lower probability surname matches could be chased over higher populations or, to the accompaniment of cries of 'what have you got to hide?', owners of rarer surnames ('Guilty', says news piece author) could be sampled nationally, or even internationally. As with most such matters, it all depends on how much you trust the people running the system. Pleasant dreams. ®

* Elimination sampling can involve pressuring the local population to 'volunteer' on the basis that those declining must have something to hide (example). Subjects may also feel pressured to allow police to keep the sample after the enquiry has concluded, and the employment of this approach in areas and populations of high crime will tend to magnify the presence of these populations on the 'voluntary' national DNA database. So you can see how the voluntary nature of the database's compilation must result in the national suspect list becoming heavily loaded against ethnic minorities and residents of poorer districts.