Search This Blog

Thursday, October 13, 2011

Amanda Knox, DNA and Circumstances

Amanda Knox, as most know, recently suffered the same fate as Casey Anthony: she was set free.
The culprit here was DNA and probably to some extent the CSI effect.

DNA identification is not the simple cut and dried thing you see on TV.

DNA is made up of a string of "values" labeled C, G, T, and A (we are talking logically here and not about the underlying chemicals and biology).  Every person has a unique string of these four values about four billion characters long that makes them unique in the DNA sense.  (Think of one giant text file full of nothing but those four letters.)

Your DNA also has structure, that is, sequences that are common to all humans as well as sub-structures within those structures.

In the case of DNA testing the model is not to compare your entire DNA what that found at the crime scene.

Instead the CSI-types look for certain markers.  A marker is a sequence of the four letters that is understood to be part of all humans.  Take D8S1179 - this marker is a series of seven to twenty TCTA sequences.

Now these sequences are individual molecules and therefore hard to detect.  What the CSI-types use is something call PCR to amplify the DNA - no all the DNA - just some pieces - like D8S1179.

D8S1179 is surrounded by markers (other unique DNA sequences).  So, in order to amplify it, first we have to find it.

So PCR is given D8S1179 plus two flanking DNA values - one for each end of the D8S1179.

PCR then searches the DNA provided from the crime scene.  When it finds a match, that is D8S1179 plus the two flankers together, it replicates what it finds doubling the amount of matched DNA.

However, the DNA from the crime scene might only be 10 ^ -12 grams (.000000000001 g) of DNA.  So you have to do the PCR process at least 28 times (in the USA) to a usable result (basically increasing the matched DNA by 2 ^ 28 for doubling 28 times or 268,435,456) which gives about 250 micrograms of matched DNA.

Now using just D8S1179 is not enough because there are only thirteen unique values for it.

So the same process has to be done for a large number of markers like D8S1179 - let's say fifty - to ensure that the markers that are unique to you are not going to match someone else.

But, as always, there are problems with this.

First and foremost is the question of the purity of the DNA collected in the first place.  In the case of hair, for example, everything fed into the PCR process for amplification might come from the hair.  But what does that mean?

First off, its why people doing this wear gloves.  If I touch a piece of someones hair with my finger my DNA might mix with the DNA in the hair.  So what would PCR be amplifying?  My DNA or the hair DNA?

PCR does not "know" what its doing - it amplifies whatever matches - whether its your DNA or mine.  PCR can just as easily amplify the wrong thing.

So as the source of the DNA becomes questionable so do the results of the test.

The longer DNA lays around the more it degrades, i.e., the base pairs break apart due to chemical reactions with the air and so on.

So given a swab of two peoples DNA and a whole slew of possible markers in the US court system at least 50 (with today's equipment) markers have to have a significant match between the person and the "crime scene" DNA for their to be a true "criminal" match made, i.e., its your DNA and not someone elses.

But, as I railed about in other recent posts, this is just a statistical result.  No one has tested all the humans on the planet and their DNA.  This is what science, through statistics, thinks is unique.  And it does not mean there wouldn't be anomalies such as someone matching you even though you are not related (or at least do not appear to be).

DNA does not provide a "timescale."

While I might find your DNA in my apartment the fact that its there at all says nothing about when it was put there, as in before or after the crime.  So there's a knife on my counter with your blood on it.  The DNA-tech goes around the house collecting up my DNA - say on a comb in the bathroom - and then touches the knife.

Did the tech get some of my DNA on the knife handle?

Was my DNA on the handle in the first place?

And on and on...

In the case of Amanda Knox there was her boyfriend's DNA on the bra clasp of the dead girl and some of her DNA on a knife handle.

The bra was not collected until six weeks after the crime.  Knox's involvement at the crime scene was linked through the supposed presence of the bra clasp DNA of her boyfriend.

Her DNA on the knife handle was minimal.

The initial Italian inquest considered that 41markers had matched for the boyfriend and Knox - not the USA required 50.  But in her first go around in the Italian justice system this convicted her.

On the second go around 41 was deemed not enough by the appeals court.

And she was set free.

Just like Casey Anthony.

Did Knox kill the girl?

No one knows.

Her African boss was also convicted of murder and remains in jail.

The problem with all of this is its not a "hard" science.  We are not blasting particles into a chamber and measuring neutron dispersion within .0001% of some mathematical model.

We have, like archeology I suppose, a bunch of fallible humans running around doing things which technology will amplify by some 2 ^ 28 in order to create a result.  A result which we could not get any other way.

Then we are doing this a few hundred times and check to see if at least 50 markers match.

Okay if nothing is screwed up along the way (and now you can see why the "chain of evidence" is so important).  If someone mixes up articles being tested for DNA, or touches them, or combines them, the whole house of cards falls apart - and there isn't even a way to tell.

Which to me is the most troubling.

PCR amplifies whatever DNA it finds - right or wrong.

It relies on people to handle 10 ^ -12 grams of material in a reliable way - something you cannot see or touch.

If you can't see or touch it, how do you know its "right"?

How do you know there wasn't contamination?

The point is you don't know.  You have to rely on statistics.  Its only the probability that its right - not that it is right.

The same kind of statistics that the Casey Anthony jury felt were insufficient to convict Anthony.

No comments:

Post a Comment