Familial DNA Testing Nabs Alleged Serial Killer

cremeglace writes "A quarter-century of conventional detective work failed to track down the killer responsible for the deaths of at least 10 young women in south Los Angeles dating back to the mid-1980s. But a discarded piece of pizza and a relatively new method of DNA testing has finally cracked the case, police announced last week. On July 7, L.A. police arrested Lonnie Franklin Jr., 57, a former garage attendant and sanitation worker they suspect is the serial killer nicknamed the 'Grim Sleeper.' The key evidence? A match between crime-scene DNA and the suspect's son, obtained by a search through the state's data bank of DNA collected from 1.3 million convicted felons."

Comments

Data mining gone wrong.

[TheTurtlesMoves]

DNA fingerprints are not as random as many think. The markers used were not designed for a nation wide database situation. Hence collisions could be a big problem. That is two people with the same fingerprint (at least at the very small parts of DNA we look at) can in fact be very likely with a database this size.

I'm not saying he is innocent, but i don't think we should jump to the conclusion that he is guilty either.

In fact we may need to use SNPs (Single nucleotide polymorphisms) to be good enough for a database of millions (or eventually billions) to reduce collisions to acceptable levels.