Re: OT: DNA Matches Jameis Winston To Sample Taken From Accu
Posted: Thu Nov 21, 2013 6:20 pm
I'll weigh in:
1) DNA breaks down in tiny fragments fairly quickly unless conditions are precisely maintained, so "compromised" DNA is incomplete DNA, so that when it is sequenced, you may find "matches" but they are not statistically unique
2) DNA is EVERYWHERE and so "tainted" means there may be external DNA added to the sample.
DNA analysis in cases like this are not full sequencing (too time consuming and very expensive). They look for specific regions of the DNA that two people are unlikely to share and only sequence those tiny portions. There are 13 regions (called STRs) in particular that are usually sequenced. There are different types of sequences that can be found at each STR and the likely-hood of any two people having the same type form is actually pretty high, about ~20%. But since there is a only 0.2^13 chance that two people will have all 13 of the same types of STR, we pretty precisely call matches. Having worked with DNA for awhile, I can tell you that the sequencing method is fairly error-proof. They use short chunks of DNA, "primers", that attach specifically to the start of these STRs and replicate only those sequences, then they read off the nucleotides of the sequence by machine. DNA absolutely won't be applied if the primers don't attach, and they virtually never attach to the wrong region, unless they aren't designed well. I see very little chance of error in the process. The error comes in when outside DNA, who might have some of the same STRs as the suspect, is inadvertently added to the DNA/STRs of the sample, incorrectly skewing the likelihood to be greater than it really is. Your argument is essentially that you may match all 13 STRs but that a contaminant added 5 STRs you don't really have to your sample...so the odds are really 1:390,625 that it's not you (instead of 1:1,220,703,125) and hey, there are way more men than that in Tallahasee (or wherever).
1) DNA breaks down in tiny fragments fairly quickly unless conditions are precisely maintained, so "compromised" DNA is incomplete DNA, so that when it is sequenced, you may find "matches" but they are not statistically unique
2) DNA is EVERYWHERE and so "tainted" means there may be external DNA added to the sample.
DNA analysis in cases like this are not full sequencing (too time consuming and very expensive). They look for specific regions of the DNA that two people are unlikely to share and only sequence those tiny portions. There are 13 regions (called STRs) in particular that are usually sequenced. There are different types of sequences that can be found at each STR and the likely-hood of any two people having the same type form is actually pretty high, about ~20%. But since there is a only 0.2^13 chance that two people will have all 13 of the same types of STR, we pretty precisely call matches. Having worked with DNA for awhile, I can tell you that the sequencing method is fairly error-proof. They use short chunks of DNA, "primers", that attach specifically to the start of these STRs and replicate only those sequences, then they read off the nucleotides of the sequence by machine. DNA absolutely won't be applied if the primers don't attach, and they virtually never attach to the wrong region, unless they aren't designed well. I see very little chance of error in the process. The error comes in when outside DNA, who might have some of the same STRs as the suspect, is inadvertently added to the DNA/STRs of the sample, incorrectly skewing the likelihood to be greater than it really is. Your argument is essentially that you may match all 13 STRs but that a contaminant added 5 STRs you don't really have to your sample...so the odds are really 1:390,625 that it's not you (instead of 1:1,220,703,125) and hey, there are way more men than that in Tallahasee (or wherever).