Julian's science

My research is in the field of the Physics of Medicine, trying to use advances in the physical sciences to lead to clinical improvements for patients. In my case, I am interested in the structure of DNA, and how genes are controlled. One of the pieces of work is to design, make and test new anti-cancer drugs that bind to DNA. I am also using nanotechnology to design DNA-based materials. One of the applications of this is designing improved components for photovoltaic solar panels.

Evidence-based policy

When making policy decisions, it is essential to make decisions based on evidence. While politics and politicians must make the final decisions - because ultimately value judgements matter - these decisions should be based on an understanding of the evidence, whether the underlying science or proper analyses of trials and statistics. The considerations that are required are becoming ever more technical, whether in fields such as health care, transport policy, education or the environment (for which please read the excellent book 'Sustainable energy wthout the hot air' Prof David MacKay FRS, with whom I have published some other work) .

However, too few active politicians, particularly in the House of Commons, have a training in and understanding of the details of science and technology, meaning that policy decisions are made without considering the evidence. This can be seen very clearly at the moment, with the firing of Prof David Nutt from the Advisory Council on the Misuse of Drugs for giving unwanted scientific advice.

The DNA database

I have become increasingly concerned with the breadth of the DNA database in the UK, and the way it is being applied. This is to me one of the critical scientific aspects of civil liberties at the moment, as well as the obvious issues of principle. I was elected to the National Council of Liberty partly with a view to strengthening their understanding and input into these issues, and have produced information for Liberty and various parliamentarians. Below I discuss some recent issues involving the DNA database, including the fact that it is now possible (even easy) to fake DNA samples.
 

 Keeping DNA data on the innocent

The UK government keeps a database of DNA samples from a wide variety of people, with over 5.6 million records. This covers both those who have been convicted of an offence, and those who were arrested but not convicted (or even necessarily charged), as well as those who provided voluntary samples (eg for elimination). Currently, all this data is kept forever. However, in the landmark case at the European Court of Human Rights, S and Marper, it was decided that data on those not convicted could not be kept permanently, as it breached article 8 of the ECHR.

The government produced a set of proposals in response to this, using highly dubious statistics to argue that they should keep data for 6 years (12 in some cases) even on those never even charged with a crime. I (and many others, including Liberty) responded in detail to this consultation, arguing that strong evidence is required to keep DNA data on the innocent, and that this data was simply not present - the consultation assumed that those who are not convicted are as likely to offend in the future as those who have been convicted, an assumption in stark contrast to all principles of the presumption of innocence. My response is available here.

The government has now announced new proposals. Despite the fact that 'the significant majority [were] opposed to any form of retention of profiles and fingerprints for persons arrested and against [whom] no further action was taken or acquitted', they have decided to keep data for 6 years on those not convicted, and are awaiting legislative time to bring these new rules in. In the meantime, data on innocent people is still being collected, with the Association of Chief Police Officers (ACPO Ltd) advising the Police to ignore the ECHR judgement. Over 90,000 innocent people have been added to the database since the ECHR judgement that it was illegal.

Does the DNA database actually work?

One fundamental question that is rarely asked is whether the DNA database actually works, even neglecting any civil liberties implications. Much to my suprise, there is very little research suggesting that it does. This is especially surprising, as I would assume that it was a useful tool. The question that needs to be asked is not just if it used, but how often it actually changes the outcome of a case. It is not useful if it merely confirms that the stereotypical person standing with a bloody dagger over the corpse and admitting a murder did indeed do the act!

There is only one statistically appropriate study like this performed, at least as reported by the government, and it was performed in the USA. (More details are in my consultation response as above). It was a relatively small sample, and in some areas studied there was no benfit at all. In other areas, there were more suspects found using the database, more arrests and more trials - but there is no data as to whether there were more convictions. We certainly do not want to use any technology to arrest more people if we're not arresting the right people! More research, and particularly more systematic and controlled research, is urgently needed to find out how useful the database could be, and hence how much its benefit compares to the necessary civil liberties cost.

From the data that is available, there are real concerns as well. According to the Police's own National DNA Database Report 2007-9, over the last two years the total number of DNA-related detections has decreased by 10,000 (25% of the original total), even though the number of people on the database has gone up by 25% over the same period. Something is clearly going wrong if having more people on the database leads to a reduced usefulness.

Faking DNA samples

DNA fingerprinting has a reputation for being extremely reliable, and has been accepted as absolute proof of guilt, even in the absence of other evidence. One example of this is R v. Adams (1997), where Adams was convicted despite the victim saying he didn't fit her description and him having an alibi. (This is also an important case for the use of Bayesian statistics in court, another of my interests, and is described in more detail here).

The reason that DNA data is treated so favourably is that it is held to be unfakable. While any jury would accept in principle that a gun, say, could have been put somewhere in an effort to frame someone, it is believed that it would not be possible, say, to leave blood stains at a scene in the same way. However, recent work by a forensic science group (Original paper available here) has shown that it is in fact fairly easy to fake DNA samples.

I have written a lay analysis of this paper (available here) to advise Parliamentarians, Liberty and others on the implications of this work. It has been used in 'DNA clinics' around the country,  In summary, however, it shows two things:

• Given a sample of DNA from an individual (such as from a hair, a cigarette butt, or saliva), it is fairly easy to create another sample, (eg blood or saliva), that will pass forensic tests as that individual.
• Given a database entry (DNA fingerprint) for an individual, this procedure can be done without any access to the individual concerned.

DNA testing and immigration

It is, of course, possible to misuse DNA testing in many ways. One recent approach was tried by the UK Borders Agency, which piloted a scheme of DNA testing of 100 asylum seekers from Somalia, to see if they really were from where they claimed. While this may sound like a good idea, the problem is that it simply doesn't work! DNA testing like this allows us to learn a lot of information about someone's ancestors, and by inference where they were from - but these ancestors are from many, many generations back. Just because someone has a great-great-great-great-great-grandfather from Kenya (and the generations are much further back than that), doesn't mean their claim to be from Somalia is false! The whole analysis also assumes that tribal groupings can be neatly divided by country, which any student of history (particularly in Africa) will realise is not the case.

Following an uproar from some scientists, the scheme has been 'temporarily suspended' by the UKBA. However, it should have never been tried in this way without being scientifically supported and tested. Instead it was pushed through with little or no scientific input. According to Nature, a prestigious if sometimes understated journal:

The border agency says that the project has undergone scientific peer review, although it is difficult to say by whom: several geneticists contacted by Nature saw a preliminary proposal from the UK government in 2007, and warned that it was unlikely to work.