DNA doesn't lie, but it can be wrong

An affirmative title ' Determining blood ties: achieving 99.9% accuracy ' probably gives you absolute confidence in DNA analysis techniques. However, this technical fact is also flawed, and sometimes it leads to the unexpected catastrophe for the person concerned.

Since DNA analysis techniques were developed in the mid-1980s, DNA testing has been widely used in clinical diagnostics and biomedical research. In addition to this " traditional " field of application, over the past 10 years, DNA testing has also been used in determining blood and forensic relationships. In the US since 1986 DNA test results have been accepted as evidence in court. After that, the courts in the world followed suit.

DNA testing has helped law enforcement agencies search for criminals effectively, and helps the court to resolve unfairly in cases of unjust charges. Few people know that it is because of DNA testing (through the saliva on the postage stamp) that the US federal investigative agency detects the "Unabomber" character, the doctor of mathematics Theodore Kaczynski, who once caused terror to the commune. American society for a while because he sent mail bombs (explosive letters) to the leadership of the industry companies he did not like. It is also the comparison of DNA from an excavated skeleton from Brazil and some family members whose historians have identified the remains of Josef Mengele, a notorious Nazi war criminal. In Germany.

However, in fact, DNA testing, like any scientific analysis, is somewhat uncertain. Never have 100% accuracy. Not all DNA evidence is considered to be the final result and the only evidence to convict someone, or to conclude a blood relationship.

Mistakes from DNA evidence

Indeed, in addition to these spectacular successes, DNA testing has contributed to serious misdemeanor decisions. In 2004, Brandon Mayfield, a practicing lawyer in Portland (USA), was imprisoned for two weeks by federal police on suspicion of a bomber on a tram in Madrid (Spain) a few months earlier. . The reason for suspicion is simple: American police found his DNA records coincided with the DNA records from the scene in Madrid. A forensic expert confirmed that coincidence is true because the probability of coincidence only occurs 1 in 200 million times. However, Spanish police insist that Brandon is not the culprit and the DNA result may be wrong. Spanish police continue to investigate and find another man with DNA records coinciding with DNA records from the scene, and through questioning, the person confessed to being the culprit. Mayfield was released, US police confessed confusingly in DNA analysis and apologized to Mayfield.

Brandon Mayfield is not a rare or unique case. In an analysis of 86 cases of oath jaw in the United States, the researchers found that 54 cases (63%) were caused by a mistake from DNA testing.

Why are there errors in DNA testing?

To understand why there are errors and uncertainties in interpreting DNA test results, it is necessary to consider through the process of applying the DNA evidence to the family.

- Step one, detect the coincidence of DNA profile (DNA profile). In this step, the authorities must prove that the DNA test results from the site coincide with the DNA test results from the suspect.

- The second step is to identify it as a true coincidence, not random. There are many reasons and the possibility of two DNA samples being the same (discussed later in this section), so another extremely important step is that authorities must prove that the DNA profile coincidence in step 1 is real.

- Step three is to identify the origin of DNA samples. Although the DNA and field polymerization from suspects is true, the suspect may not be the culprit, but just " accidentally " leaving a trail at the scene. Therefore, authorities must prove that suspects are present at the scene;

- Step four is the verdict. After all of the above information has been confirmed, the court or new jury has reason to decide whether the suspect is the culprit or innocent.

Each step and each sequence of links between the two steps in the above procedure can be flawed. If errors occur from the first step then all the results and information of the following two steps become meaningless, and the evidence is not accepted in court.

In step one, DNA test results can violate technical errors such as contaminated or damaged enzymes, samples (samples of blood, hair, saliva .) or salt concentrations for DNA analysis. abnormal, or due to confusion of blood samples, or simply a technician's mistake. It is very difficult to know what the error rate is in step one (because few people make a technical error), but usually ranges from 1 to 5%. Through reevaluation 75 reports of DNA duplication, it was found that there were 3 errors in step 1, ie 4%.

Mistakes from interpreting probabilities

If step one has no flaws, and authorities detect a coincidence, the problem is how much is this probability of coincidence? To answer this question, authorities must apply probability theory to estimate random match probability (RMP).

To understand this probability, it is necessary to consider the following example. Suppose the experts document DNA by analyzing 6 genes (each gene has many variations, ie genotype), and the result is a variation of 6 genes analyzed from samples taken from the field and samples taken from human blood. completely suspected. But can it be concluded that the suspect is the person who left the blood sample at the scene? The answer is " no ", because there may be other people who have the same gene variant as the suspect: DNA records only analyze over 6 genes, but in the human body there are more than 30,000 genes.

Thus, even if the DNA test results are completely accurate, the interpretation of DNA results may be wrong because the interpreter does not understand what the DNA polymerization means. The probability of DNA polymerization such as 1 in 10 million, or even 1 in 57 billion (as is the case with OJ Simpson) does not mean the probability of an innocent accused. DNA test results can only be considered as a source of evidence in many other evidence, and need to be interpreted in accordance with its uncertain meaning. That requires a strong knowledge of probability theory and demographic genetics.