CRIME
The Truth About Genetic Evidence in Forensic Cases
Sat Jun 07 2025
Forensic science is getting a boost from new sequencing technology. This tech lets scientists get genetic info from tough samples. One new tool, IBDGem, looks at sequencing data, even from low-quality samples, to figure out if a DNA match is likely. But here is where things get tricky. IBDGem checks if the sample comes from someone in a reference database. This is not usually what forensic experts want to know. They typically want to know if the sample matches a suspect, not someone in a database.
The likelihood ratios that IBDGem produces can be misleading. They can seem to show a strong match when there isn't one. This is because IBDGem compares the sample to a database, not to a specific suspect. This can give a false sense of certainty. For example, if the sequencing data has no errors and doesn't match anyone in the database, the likelihood ratio becomes infinite. This doesn't mean the match is certain; it just means the math breaks down.
So, what's the fix? Scientists need to find a way to make likelihood ratios that test the usual defense hypothesis. This means comparing the sample to a suspect, not a database. Until then, forensic experts need to be careful. They should not rely too heavily on IBDGem's results. They should also consider other evidence and context.
It's also important to note that sequencing errors can affect the results. Even small errors can change the likelihood ratio. This is why it's crucial to have high-quality sequencing data. It's also why forensic experts should always question the results. They should never take them at face value.
In the end, the goal is to use genetic evidence fairly and accurately. This means understanding the tools and their limits. It means questioning the results and considering all the evidence. It means making sure that justice is served, not just math.
continue reading...
questions
How can forensic scientists ensure that new technologies like IBDGem are rigorously tested before widespread use?
What are the potential biases in the reference database that could affect IBDGem's results?
If the sequencing error rate is zero, does that mean the sample is from a perfect human or a robot?
inspired by
actions
flag content