Family Ties to Crime

After commenting rather heavily on the math in DNA matching and the utility of trolling through a criminal DNA database, I find this piece in Reason Magazine:

The BTK murderer in Kansas was identified using DNA from a pap smear that his daughter had when she was in college. The police checked her DNA against DNA collected from the crime scenes and it was a close match. With this evidence, the police got a warrant requiring the suspect to submit to DNA testing and it matched perfectly. The murderer is now serving ten consecutive life sentences.

As the article points out, such familial DNA searching causes unease among civil libertarians. To wit:

"If practiced routinely, we would be subjecting hundreds of thousands of innocent people who happen to be relatives of individuals in the FBI database to lifelong genetic surveillance," said Tania Simoncelli, science adviser to the American Civil Liberties Union....

The Supreme Court has repeatedly held that authorities may not conduct searches for general law enforcement purposes without suspicion about individuals. Although convicted criminals have a diminished expectation of privacy, searching a database for unknown relatives might violate that principle, said Jeffrey Rosen, a George Washington University law professor.

"The idea of holding people responsible for who they are rather than what they've done could challenge deep American principles of privacy and equality," he said. "Although the legal issues aren't clear, the moral ones are vexing." ...

Stanford University law professor Henry T. Greely estimates that at least 40 percent of the FBI database is African American, though they make up only 13 percent of the U.S. population. That is because in an average year, more than 40 percent of people convicted of felonies in the United States are African American, he said.

If the national database were used for familial searching, he said, and assuming that on average each person whose profile in the database has five first-degree relatives, authorities would be "putting under surveillance" roughly a third of the African American population, compared with about 7.5 percent of the European American population, he said.

"I don't think anybody's going to be falsely convicted," he said. "It's the time, hassle and indignity of being interviewed by the police. How much is that worth? How much does that cost a person? I don't know, but it's not zero."

The general principle behind familial DNA matching is pretty simple.

DNA testing looks for markers in DNA, which are distributed pretty randomly through the population. Any given marker will show up in a small percentage of the population, usually between 5 and 25%. Thirteen markers are commonly used for a genetic ID.

If the DNA that's compared belongs to someone who is a close relative of a suspect, we'd expect to find a different number of markers in common -- say, 6 or 7 for a parent, child, or sibling; 3 or 4 for a grandparent, grandchild, uncle, aunt, niece, or nephew.

Of course, at two degrees of separation, like the uncle or nephew relationship, we're looking at pretty shaky connections. But the 6 or 7 marker match might lead us to look at close relatives of the person whose DNA was matched. (Indeed, it's a recurring plot element in episodes of CSI.)

What are the ethical implications of, say, trolling through a database of criminals, and looking for relatives of anyone with a 6 or 7 marker match? How about trolling through a bone marrow bank's records and looking for exact or family matches with donors? (My HLA profile is on record with the Red Cross.)