Slashdot Mirror

Good find. The correct link is: Science Letters - Time for DNA Disclosure

Except that the link has 404'd. Wandering around the website, I found this link which appears to be a number of articles on reviewing forensic DNA testing literature. More stuff to read....

Letter is at http://www.bioforensics.com/articles/Krane_Science_letter_2009.pdf/

The central issue is how a DNA database match should be weighed in court. The 1 in 113 billion figure is the chance of randomly selecting a single unrelated person from the population with a given DNA profile. A database search is not a random selection and a database likely contains relatives. The statistics that are currently in use assume the individuals being searched are unrelated. Some labs will factor in the size of the database (np method), while others will not.

A couple of corrections with the above comments:

The term "DNA Fingerprinting" is quickly disappearing and was argued against from the beginning by many due to the lack of a sound scientific foundation for fingerprints. "DNA Profiling" is much more appropriate.

The Arizona database search utilized their state database of approximately 65,000 individuals, not 6 million. It is expected that you would see around 100 pairs of matching individuals at 9+ loci when the profiles are unrelated. That means that there are around 44 additional pairs of matching individuals. The most likely cause is the presence of relatives (approximately 1000 sibling pairs). (see PPT Presentation)

Finally, these issues are further magnified when labs perform familial searching. Sometimes a database search will produce a close, but imperfect match to a database entry. The partial match will sometimes be used as justification to investigate a relative of the person in the database. This introduces many issues with what constitutes probable cause and civil rights. It's also important to have a sound statistical method for identifying when the true perpetrator is likely to be a relative (see Paoletti, et al. 2006).

The central issue is how a DNA database match should be weighed in court. The 1 in 113 billion figure is the chance of randomly selecting a single unrelated person from the population with a given DNA profile. A database search is not a random selection and a database likely contains relatives. The statistics that are currently in use assume the individuals being searched are unrelated. Some labs will factor in the size of the database (np method), while others will not.

A couple of corrections with the above comments:

The term "DNA Fingerprinting" is quickly disappearing and was argued against from the beginning by many due to the lack of a sound scientific foundation for fingerprints. "DNA Profiling" is much more appropriate.

The Arizona database search utilized their state database of approximately 65,000 individuals, not 6 million. It is expected that you would see around 100 pairs of matching individuals at 9+ loci when the profiles are unrelated. That means that there are around 44 additional pairs of matching individuals. The most likely cause is the presence of relatives (approximately 1000 sibling pairs). (see PPT Presentation)

Finally, these issues are further magnified when labs perform familial searching. Sometimes a database search will produce a close, but imperfect match to a database entry. The partial match will sometimes be used as justification to investigate a relative of the person in the database. This introduces many issues with what constitutes probable cause and civil rights. It's also important to have a sound statistical method for identifying when the true perpetrator is likely to be a relative (see Paoletti, et al. 2006).

I have never been indicted nor found guilty of any crime and as such there is no reason for the government to retain such information.