[Back to Document View]

                            LexisNexisª Academic

  Copyright 2002 Reed Business Information UK, a division of Reed Elsevier,
                                    Inc.
                           All Rights Reserved.
                               New Scientist

                               July 20, 2002

SECTION: Features; ISSN. 0262 4079; Vol. 175; Pg. 3434

LENGTH: 2649 words

HEADLINE: Ready for your close-up?

BYLINE: Clare WilsonWorking out what someone looks like from only a
DNAsample is no longer science fiction. You'd be surprised what forensics
experts can already do, says Clare Wilson

BODY:

IT'S THE police's worst nightmare. A rapist murders his victim and leaves no
clues to his identity. A semen sample from the dead woman doesn't match
anything in the forensic database, and a mass screening of local men draws a
blank. Stalemate.

But what if there were another lead for the police to follow up? Suppose a
lab analysis of the semen sample could produce a "DNA photofit" of the
suspect: Caucasian, with red hair and blue eyes, a strong jaw and a cleft
chin. Not exactly a smoking gun, but combine it with a bit of traditional
police work, and a description like that could make all the difference.

At the moment, it's not possible to generate a complete description from a
scrap of DNA. But researchers are getting there, albeit in small steps.
Already they can interrogate DNA for its owner's ethnic appearance, hair
colour and eye colour, and research groups around the world are making
progress on other physical traits, including jaw shape.Magazine Not everyone
is happy with the idea. Some civil libertarians say the strategy is an
invasion of privacy and could lead to miscarriages of justice. And some
scientists say the complexity of the task is so great that the technique
won't go much further. For a start, there might be millions of different
genetic combinations that all lead to just a handful of physical types. But
the British police are undeterred. DNA photofits have already helped them
solve crimes.

At the moment, if DNA is found at the scene of a crime, in blood, hair or
semen, say, standard forensic techniques can help in two ways. If there's a
suspect in custody the police can see if their DNA matches the "crime
stain", as it is called. Or in the absence of a suspect, they can see if it
matches the DNA of any known criminal, held in archives such as Britain's
ever-growing National DNA Database. This database holds a collection of DNA
samples taken from suspects the police have arrested and is already the
world's largest forensic DNA archive.

Both techniques have proved their worth in criminal investigations. But what
if there's no suspect and no match in the archive? Ever since DNA testing
was introduced, forensic scientists have wondered how much information they
could wring out of a DNA sample.

At the forefront of this field is Britain's Forensic Science Service, a
government agency based in Birmingham that is the custodian of the National
DNA Database. The FSS can already give the police clues about two aspects of
someone's appearance - ethnicity and hair colour - and is funding research
that could generate more detailed information from DNA. And the FSS isn't
alone. Research groups around the world are making progress with the
genetics of physical appearance, and the biotech industry is getting in on
the act too.

The biggest success story so far has been hair colour. Back in 1995, a team
at Newcastle University reported the discovery of a genetic trait linked to
red hair. They noticed that redheads usually have a faulty receptor protein
on the surface of pigment-producing cells called melanocytes. The defect
renders the cells insensitive to a hormone that instructs them to make
melanin. By examining the gene for the receptor protein, they established
that the defect is a recessive genetic trait: 70 per cent of redheads have
two copies of a defective gene (New Scientist, 30 September 1995, p 18).

The FSS has now capitalised on that result. Last year a team led by Penny
Noake announced it had identified eight common mutations that cause the gene
to be defective. More significantly, the team had also devised a test to
pick up those mutations in crime-stain DNA. The test isn't perfect, but
based on a sample of 197 Caucasians, the FSS concludes that someone who
carries two of the defective genes has a 96 per cent probability of being
naturally red-haired.

It's not hard to see how the test could be useful to an inquiry. If the FSS
tested a stain and found the signature for red hair, the police would know
to concentrate their investigation on suspects with red hair - allowing for
baldness, greyness, dye and so on. And a negative result would help them
eliminate red-headed suspects. Importantly, the test result doesn't have to
stand up in court - its only purpose is to help the police direct their
investigation, not convict someone. Once they've brought someone in for
questioning they can take a cheek swab and check the suspect's DNA against
the crime stain.

However, the red-hair test is of limited use. Only 6 per cent of people
living in Britain are red-headed. What about blonde, brown and black-haired
criminals? Unfortunately, it's at this point that the genetics of hair
colouring spirals into bewildering complexity. Hair colour is usually
determined by the cumulative effect of several genes, so there's no such
thing as a single gene for blonde hair that could be turned into a simple
test, for example. It's the same with eye colour.

But biotech firm DNAPrint Genomics of Sarasota, Florida, is having a crack
at both problems, and has made most progress so far with eye colour. Its
starting point was basic research in mice and the fruit fly Drosophila that
identified several genes involved in eye colour. Geneticists then looked for
similar sequences in the human genome, and identified 10 candidate genes for
eye colour.

DNAPrint has now taken that information and turned it into a test. The firm
took DNA samples from 500 volunteers and recorded their eye colour. Then it
applied a technique called SNP mapping to find correlations between the two.

SNP stands for "single nucleotide polymorphism" - a single "letter" change
in the genetic code. These variations account for most of the genetic
differences between individuals. DNAPrint sequenced the 10 candidate genes
from each volunteer then sifted through the sequences looking for SNPs. It
found 50 in total. Then it set computers to work out how the SNPs correlated
with eye colour.

Of the 10 genes, DNAPrint found that only four really matter. Chief
executive Tony Frudakis says that by looking at these four genes, it can
classify someone into one of three eye-colour groups - dark (black and
brown), light (blue and grey), or hazel - with 97 per cent certainty.
Exactly what the four genes do, and how they combine to determine eye colour
remains a mystery, but that's not a problem for DNAPrint: to make a
commercial test it doesn't need to know how it works, just that it does.

The next challenge is hair colour. DNAPrint is applying exactly the same
technique to this problem, identifying candidate genes and looking for SNPs.
Frudakis says they have made some headway and can tentatively classify
people as blonde, red, auburn, brown, or black-haired from their DNA alone.

Back in Britain, another aspect of appearance the FSS has had success with
is ethnic origin. Its starting point was the National DNA Database. Since
1995, the FSS has analysed and classified the DNA samples in a standard way.
It creates a genetic fingerprint for each, using 10 regions of non-coding or
"junk" DNA that are known to exist in hundreds of different variants, known
as alleles. The resulting fingerprints are useful in forensics because
almost everybody has a unique combination. If the fingerprints of two
samples match then it's overwhelmingly likely that they come from the same
person.

In the mid-1990s, FSS researchers began to wonder if they could winkle out
any other information from the fingerprints. One promising idea came from a
piece of supplementary information recorded by police whenever they take a
DNA sample: the person's ethnic origin, either Caucasian, Afro-Caribbean,
South-East Asian, Indian sub-continental, or Middle Eastern.

In March 1996, the FSS started analysing the entire National DNA Database to
see how often each allele cropped up in each ethnic group. They found that
some variations were much more common in certain groups than in others,
suggesting that crime stains could provide clues to suspects' ethnic origin.
Then, last year, the FSS announced that it had made this idea into a
reality: it can now estimate the relative probability that a crime stain
came from a member of any given ethnic group.

The genetics of ethnicity, of course, is a controversial subject, with most
scientists now agreed that conventional racial categories don't correspond
to genuine genetic groupings (New Scientist, 20 April, p 34). But the FSS
test doesn't contradict this consensus, for two good reasons. First, the DNA
sequences involved are junk rather than functional genes, so they're not
genes "for" ethnicity. And second, the ethnic category assigned to DNA
samples is purely subjective, based on the donor's appearance. It may not
tell the police anything meaningful about someone's "race", but it will tell
them what they look like, which is what matters for these purposes.

The ethnicity test has already been put through its paces in real
investigations. Andrew Urquhart, a geneticist at the FSS, says it has been
most useful in cases where there was no eyewitness evidence. "Sometimes the
police have hundreds of people to screen on a case," he says. "We can say:
'Screen these people first'."

Although the FSS won't discuss individual investigations, a case study
published in the journal Forensic Science International (vol 119, p 17)
demonstrates one way in which the ethnicity test can help. A young woman was
raped by a man she described as Afro-Caribbean. The police took a swab and
profiled the DNA in the semen, but they couldn't be sure it was the rapist's
DNA because the victim had recently had unprotected sex with a Caucasian man
who couldn't be traced. Was the semen from an Afro-Caribbean man? That is,
was it worth taking DNA samples from suspects to match against the semen?

The ethnicity test helped answer the question. The allele combination in the
semen would be expected to occur in 1 in 3 billion Caucasians, but only 1 in
83 billion Afro-Caribbeans. In other words, the semen was more than 27 times
as likely to have come from the victim's Caucasian partner than the rapist.
The police decided against using DNA matching.

Beyond hair and eye colour and ethnic appearance, divining physical
characteristics from DNA gets more difficult. Even so, the FSS has also been
pursuing the genetic basis of facial features. A few years ago it helped
fund a large data-collecting exercise at University College London. Over
several months, an exhibit at London's Science Museum invited visitors to
leave DNA samples and have their faces scanned using 3D surface mapping.
About 600 people volunteered, many from the same families.

The researchers tried to break down overall facial shape into distinct
features such as nose curvature or chin clefts, and correlate them to DNA
sequences. But they made little progress. Just as with eye colour, there is
no one gene for a big nose, so the enormous complexity of the task defeated
the researchers. They did identify some candidate genes for cleft chins
though, but when the lead scientist retired the project was wound down
without drawing any firm conclusions.

Alf Linney, reader in medical physics at UCL who was involved in the
project, recalls: "The Home Office had great confidence we could do it in a
couple of years, but we didn't. There are many genes involved in some of the
features. The magnitude of the problem was too great."

But the idea of finding genes for facial features isn't dead. Many of the
genes involved in development, and therefore in physical appearance, are
common to most mammals. So a gene for a large jaw in mice, for example,
might very well be found in humans too. And working with mice has big
advantages, since they can go through many generations in a short time,
which is extremely helpful in genetic studies. What's more, bone shape can
be determined very accurately once the animals are dead.

One promising project has found that mice show significant variation in jaw
shape and size, and has begun to unravel the genetics behind the variation.
Project leader Chris Klingenberg of the University of Konstanz in Germany
cautions that, as with humans, the genetics controlling jaw shape in mice is
fiendishly complicated. "There are amazingly many interactions," he says.
"If you change gene A, genes B and C interact in a different way with each
other. And this is in a system picked for its simplicity."

But the project is making some progress anyway. In one study of 535 mice it
turned up 25 genes for jaw shape, another 12 for jaw size, and five more
involved in jaw symmetry. And despite the mind-boggling number of possible
genetic permutations this throws up, the number of physical outcomes is
surprisingly small - Klingenberg has found only two basic patterns.

Klingenberg says that the ultimate goal of applying this information to
human faces is still a long way off. But when clues to the genetic
mechanisms do emerge, the UCL work on facial features may not be in vain:
their DNA samples and face scans have been archived and could be used in
future research, once scientists know which genes to look for.

With physical characteristics, however, genes never tell the whole story -
environment plays a key role too. FSS researcher Kevin Sullivan points out
that when it comes to someone's facial characteristics, "whether or not
they've played rugby might have more of an influence on ear and nose shape
than genes."

Urquhart agrees: "There are an awful lot of caveats. Someone could be bald
or dye his hair. Somebody might have the genes to be tall but suppose
they're malnourished as a kid."

These are problems that can be taken into account, but there are other, more
serious objections. Alec Jeffries of Leicester University, who invented
genetic fingerprinting, points out that the tests may violate one of the
principles of forensic testing, namely that police should not be able to
find out personal or medical information from DNA. He says that many genes
involved in physical appearance are also connected with inherited diseases.
For example, pigmentation genes are involved in skin cancers, and mutated
versions of facial genes could cause congenital abnormalities.

"The police have absolutely no right to be gaining that sort of information
from the genome," Jeffries says. "One only needs to go back in history to
realise the perversions of justice that could be caused. For example,
there's been a lot of work trying to pin down sexual orientation - the "gay"
genes. One could imagine that that might be quite a useful lead for the
police. But is it right for the police to go rummaging in DNA to find out
things like that? It's invasion of genetic privacy."

Jeffries is also unconvinced by the science. "A cynic would say it's science
fiction and it isn't going to work," he says. "With the possible exception
of hair colour, the genetic basis of all these traits is simply not
understood."

Civil liberties group Liberty also has concerns. Spokesman Roger Bingham
says: "If it enables them to build up a picture of someone's appearance as
an investigative tool, then that's not necessarily a problem. But at the
moment, the tests aren't remotely conclusive." His fear is that the police
would clutch at straws and bring in suspects based on very flimsy evidence.

Perhaps unsurprisingly, the FSS is scathing about the naysayers. Urquhart
says: "We are just finding a quick way of getting what normally comes
through eyewitness information. I don't really see a difference, except this
sort of information might be more reliable. I think if it's used in the
investigation of a crime, and it's not giving any evidence that's irrelevant
to the crime, I have got no objections." His colleague Kevin Sullivan adds:
"Law-abiding citizens don't have anything to worry about. But criminals do."

LOAD-DATE: July 31, 2002