so what's in that 0.1%?

June 5, 2003 12:30 AM

DNA used to ascertain race of unidentified serial killer. Florida company DNAPrint Genomics claims their test can identify the race (ie, African, Caucasian, East Asian or American Indian) of a person from their DNA. CEO Tony Frudakis says that "of over 2,200 blind samples tested, the test is yet to get one wrong."
posted by shoos (12 comments total)

It's not clear, however, how they judge whether or not a result is "wrong."
posted by shoos at 12:33 AM on June 5, 2003

.... come on, where are all the alarmists?
posted by Keyser Soze at 2:45 AM on June 5, 2003

What's the problem?

Or do you doubt a witness if he/she says "I saw the man wot done it... 'e was a black man!". It's selecting with race as a criteria but how's that any different to saying "He's tall"?

Give it a few years and they'll be able to show you the guy's face and replicate his voice using a dna sample...
posted by twine42 at 2:53 AM on June 5, 2003

Race, how many other countries besides the US seperate folks by race?
Ps, usually when I use the word folks, it means like me, people are others.
posted by thomcatspike at 5:07 AM on June 5, 2003

Missed it every time in a spelling bee...separate.
posted by thomcatspike at 5:10 AM on June 5, 2003

A couple things are unclear to me:

Wrong/Right seems to depend on ancestry not on skin color. This may seem like a quibble but people of similar ancestry (even siblings) may have wildly varying skin colors.

Only four distinguishable "races" seem to be in play here: (sub-saharan) Africans, (presumably Northern) Europeans, (Southeast?) Asians and Native Americans. But, what about that large swath of Northern Africans, Mediterranean, Middle East, Pacific Islander, Australian Aboriginal etc. people.

So, twine42, I believe they consider their test "correct" in identifying someone as "black" if the person is of mostly sub-saharan African descent regardless of their actual skin color. A subtle but important distinction.

Please correct me if I have misinterpreted what i have read.
posted by vacapinta at 5:29 AM on June 5, 2003

Australian Aboriginal
Saw a piece on Discovery using dna they followed the migration of Australian Aborigines to Europe then on and on, to a group of Inuit in Alaska, then finally to American Natives in Arizona.(can't recall which tribe)
posted by thomcatspike at 6:00 AM on June 5, 2003

Have they tried this on Vin Diesel yet?
posted by gimonca at 7:07 AM on June 5, 2003

I am still slightly confused how exactly this company plans on making a full physical profile from the DNA left at the crime scene which the article alludes to. As I recall from my studying of genetics, one of the reasons physical profiling isn't used not only because of scientific limitations in the past, but because most of the time there was simply not enough DNA left at the crime scene (as in there was simply not enough sequences even after the DNA was multiplied), and even the DNA that was left there would usually denature some to the point where it could throw up an entire genoming structure for just this kind of stuff. (btw, in cases, DNA finger-printing as its been done in the past isn't done by comparing actual DNA sequences, but rather by comparing the lengths of introns- useless pieces of DNA we all have that vary from person to person)
posted by jmd82 at 7:49 AM on June 5, 2003

Have they tried this on Vin Diesel yet?

Or how about that famous "Cablinasian" Tiger Woods who includes European, African, Native American, Thai and Chinese elements in his background. Should be interesting.
posted by Mitheral at 9:19 AM on June 5, 2003

Since all human beings have virtually identical DNA, geneticists have to look for slight chemical variations that distinguish one population from another. One technique involves the use of "microsatellites" - short repetitive fragments of DNA whose patterns of variation differ among populations. Because microsatellites are passed from generation to generation and have a high mutation rate, they are a useful tool for estimating when two populations diverged.

Depending on their database-they may have 21-60 distinct human populations keyed in to compare a given sample's markers against. Thus:

"Dr. Shriver said he had identified markers, meaning specific units or short sequences of DNA, that are more often found in people from one continent than another. If enough markers are used, a researcher can say with reasonably high confidence to which of the major continental races — African, Caucasian, East Asian or American Indian — a person belongs, as well as the percentage of each ancestry in the case of people of mixed race."

So he's using microsatellites-he has total confidence he can name the continental race. With enough samples-you could get it down to Iroquois or Crow.
posted by quercus at 12:52 PM on June 5, 2003

I'm under the impression that the test is based on single nucleotide polymorphisms.

jmd82: all they need is essentially a single copy of the genome. DNA is actually pretty stable, and just a milligram of your average tissue contains about a million copies.
posted by shoos at 5:09 PM on June 5, 2003