By P Koupparis
19 October 2002

There is one type of modern evidence that is always presented (and challenged) by forensic experts: DNA profiling evidence of identification. It may be thought that such evidence is so specialised as to require the services of scientific experts to assist the jury. That may be so, but that is not the only reason.

DNA profiling evidence happens to fall into a very special category: the original exhibit is too small to be seen even with the most powerful optical microscope. No juror has ever seen the evidence he or she thought was sufficient to justify a guilty verdict. The exhibits that underlie a DNA profile belong to the sub-microscopic realm of atoms and molecules.

The invisibility of profiling exhibits is not widely publicised by those on the enforcement side of the criminal justice system. Expert witnesses are employed, in part, to guard against the possibility that jurors might otherwise seek to examine the exhibits for themselves, which would be quite impossible. But size is not the only thing that makes DNA profiling evidence unique.

Since DNA profiling evidence can only 'exclude' a suspect with absolute certainty, it is invariably the case that legal arguments revolve around the significance of the statistical probabilities with which suspects are 'included' by it. The intricacies and vagaries that follow are, to my knowledge, almost never raised before a jury, for fairly obvious reasons.

We need to delve a little deeper into the technology of DNA profiling while, where possible, avoiding the mind-numbing terminology and impenetrable jargon normally associated with this subject.

Ten pairs of numbers in the UK (thirteen pairs in the US) are used to represent an individual's DNA profile. Each pair identifies the lengths of the two alleles found at each DNA locus known as a Short Tandem Repeat (STR). The problem faced by prosecutors is to link those neat, dry numbers with traces of body fluids or debris found at a crime scene.

Scene-of-crime DNA evidence is usually produced in the form of a printout either from a capillary electrophoresis (CE) machine or a laser scan of an electrophoresis slab gel. Both methods purport to identify the lengths of detected allele molecules. The CE graph displays them as peaks while the slab gel shows them as bands. Slab gels predominate in forensic criminal investigations because they can be dried and preserved for posterity.

The important point here is that the peaks or bands on those printouts define the numbers that make up a DNA profile. That is where the numbers come from. What jurors are not told is the bizarrely tenuous relationship between those numbers and the actual DNA they are supposed to be measurements of when profiling is carried out with a multiplex Polymerase Chain Reaction (PCR) kit, as almost all are, these days.

The original genome DNA exhibit (called template) is not consumed by the PCR reaction and can be recovered intact afterwards. None of the original template DNA is measured directly during the measurement process. So what is actually measured?

Multiplex PCR kits contain artificial DNA molecules called primers that are up to thirty base pairs in length. The cocktail also contains a huge number of artificially created A, C, T and G nucleotide bases (the basic building blocks of DNA).

Each profiling locus requires a forward and reverse primer pair, each of which binds to the 'phosphate' end of the sense and anti-sense strand of the template DNA, respectively, just ahead of the sequence to be amplified (in this case a particular STR locus).

The reaction takes place in a thin-walled plastic vial that holds a few drops of PCR cocktail (typically 20-50 µL) to which is added anything between one and five hundred genome DNA templates. The vial is placed in a machine called a thermal cycler that repeatedly raises and lowers the temperature between three levels.

The cocktail is first heated to 94 ºC. At that temperature double-stranded DNA separates into single strands. It also activates the modified 'hot start' polymerase enzyme Taq Gold. (Taq is the DNA polymerase enzyme of the hot spring bacterium: Thermus aquaticus. It drives the PCR as most other versions decompose at such temperatures.)

At the next temperature level, 60 ºC, the primers bind to their targets by matching bases A with T and G with C, thus forming a short segment of complementarily-paired, double stranded DNA.

Finally, at 72 ºC, DNA polymerase extends each bound primer by clipping together free-floating nucleotide bases to complete a new, complementary strand of DNA. That three-stage temperature cycle is repeated 25 to 35 times over a three-hour period.

The first two cycles yield random length DNA fragments that are direct copies of the target region of the template DNA. By the third cycle, copies of the target alleles are fully delineated by the forward and reverse primers. The targets include the primer sequences. From that point onwards polymerase preferentially copies the shorter molecular copies rather than the massive template DNA molecule(s). Each thermal cycle doubles the number of allele molecules in the vial - hence 'chain reaction'. By the end of 30 cycles over a billion copies of each target sequence have been produced.

The last thermal cycle is held at 60 ºC for 30-45 minutes to enable Taq polymerase to add an extra A-type nucleotide base to the 'sugar' end of every allele molecule, which makes them one base longer. That extra base is not part of the original template DNA's sequence but is forced in order to normalise the effect of an intermittent polymerase copying error.

The molecular soup resulting from a successful PCR amplification is called an amplicon. The final allele molecules are made entirely from primers and bases that were put there by the manufacturer of the kit. They bear no physical relationship to the original template DNA apart from being some thirty times removed from the initial, direct complementary copy of its base sequence. A PCR generated DNA profile is, quite literally, a fabricated exhibit.

To further complicate matters, standard UK and US profiling loci are subject to another polymerase copying error wherein as many as 15% of the copied molecules are exactly one repeat length shorter than they are supposed to be. The effect is called stutter and is believed to be caused by slippage over the tetra-nucleotide repeat sequence during polymerase primer extension.

A difference of just one repeat length at one allele is the difference between guilt and innocence. In a typical PCR amplicon up to one molecule in six could be interpreted as belonging to individuals other than the suspect. Without a court-approved expert acting as an intermediary, it is highly unlikely that anybody would be convicted on DNA evidence: clued-up defendants would undoubtedly exploit the infinite possibilities of innocence provided by the PCR's stutter products.

Another (quite surprising) component included in a PCR kit is an allelic ladder. The steps of that ladder are made up of human DNA fragments collected from hundreds or even thousands of people as examples of the more commonly observed allele lengths. The allelic ladder is provided to determine a) the length of a suspect's alleles by comparison and b) that a PCR batch is working properly. The latest US CODIS PCR kit has over two hundred alleles in its ladder while a popular kit specific to the UK's profiling STRs has 159.

It seems that juries are simply not told that a suspect's DNA has been amplified in a cocktail that also includes pieces of DNA taken from hundreds of other people. The allelic ladder alleles are tagged with a fluorescent red dye. The primers that eventually become the suspect's target alleles are tagged with one or other of a blue, green or yellow fluorescent dye. The manufacturers' protocols warn users that those dyes can sometimes fall off the molecules they are bound to. If I were a prosecutor, I would not want to mention any of that to a jury, either.

It is the PCR amplicon, a mixture of fabricated and pre-dyed allele molecules along with copying errors, that is separated and detected by the capillary or slab gel electrophoresis techniques. The vital profiling numbers are deduced from the intensity of light given off by the primer- and ladder-tagging dyes as their carrier molecules are dragged through a molecular sieve by a high voltage. It is only the amplitude, colour and the relative order in which the dye-emitted light blips appear at the photo-detector that serves to identify particular alleles.

After all that chemical magic, the final DNA profile is the subjective interpretation of those blips by two or more human operators. Their 'expert' opinions are based on an awful lot of assumptions about an indirectly glimpsed, invisible exhibit. No other form of physical evidence is presented so obliquely, is subjected to such a complex fabrication process or requires so much trust in its integrity.

Science it may be, but smoke and mirrors are not excluded.

© Copyright P Koupparis 2002. All Rights Reserved.

A comprehensive web resource for PCR/STR technology is at: http://www.cstl.nist.gov/biotech/strbase/

STR reference book: Forensic DNA Typing

DNA Evidence: is it safe to convict?