Inspired by an endless thread about the murder of JonBenet Ramsey.
Supposedly, ten years after the murder was committed, police discovered trace amounts of unknown DNA on her underwear. How likely is it that this DNA came from the killer? Or put another way – is the DNA testing so sensitive that it could have picked up trace amounts from someone who merely handled the laundry days before, or used a toilet seat before she sat down on it, or something like that?
You have two questions: how accurate is it, and how sensitive is it?
The basic idea is that only a handful of skin cells–fewer than a dozen, perhaps–which are shed all the time, are required to get a DNA result. The DNA from those cells can be amplified and a result given that is reasonably accurate in giving a DNA profile of the donor. So in that sense it’s sensitive and potentially accurate.
The question for a given circumstance is the context, though. Yes, it is sensitive enough that cells from a factory worker or anyone else handling the article of her clothing can be the donor of the DNA. It’s not out of the question that cells from me shaking your hand can wind up on my clothes, or clothes I handle. So it’s not “accurate” in determining who the perp is unless you have a candidate perp. Then you can place them at the scene (or at least, place them in context of having handled the clothing, etc).
In the Jon Benet case, I’ve never seen an expert analysis of how all the DNA was obtained and handled. Certainly DNA in cells on the underwear could easily cross-contaminate other articles of clothing, and it’s my opinion based on what I’ve read that the touch DNA analysis in her case is nothing close to exculpatory for suspects in the family circle. Mary Lacey, the DA who pronounced her family cleared based on the DNA, did not have sound reasoning in my opinion, and in fact had political and perhaps personal reasons for jumping to that pronouncement.
You could have a crime technician walk around your home and find DNA from dozens of sources. We shed skin cells all the time.
I suppose too it depends what sort of sample - the information we’re missing. If it’s a tiny chunk of skin cells, then who did the laundry in the househld, and could it have transferred from other items during washing, etc.? If it’s a dried stain of something else, it’s a bit harder to suggest it came from accidental contamination.
Obviously, DNA, like every other piece of evidence, just lends context. It does not prove anything. The circumstances surrounding where it was and how it could have gotten there are always the critical issues.
IIRC, there is an “amplification” process which can create billions of copies of a tiny sample for analysis. In another scenario, a perp sample with a tiny flake of cross-contamination from someone else might yield a debatable result. Then, as pointedout in another thread recently, a small sample may not produce enough information to positively identify anyone. In the ideal case, with a nice clean large sample, the odds of misidentification are pretty much zero - one in billions, or something.
There was an episode of CSI:NY (not the best source for accurate information) where they thought they had a serial killer, only to find that a person at the q-tip factory was not wearing her gloves while handling the long yarn of cotton being made into q-tips; she was leaving a residue of skin cells every so often. Labs all across the north-east cameup with a DNA match for the same prson in several different homicides. Sounds a little far fetched, but not impossible.
I just happened to watch an episode of Cold Case Files that discussed this…
The victim was a young woman who was raped and murdered in Ann Arbor, Mi, in 1969. Many years later, the investigators were able to generate DNA profiles from several stains on her stockings and one single blood stain that was found on her.
They convicted one man based on the sweat stains; presumably the size of the stains precluded any accidental transfer.
Here’s the surprising bit: the single drop of blood found on the victim was traced to a prison inmate who was a small child at the time of the murder. They questioned him, and though he could recall some dark events in his past, they couldn’t figure how his blood came to rest on the girl’s body.
The defense did research and found that a sample of the inmate’s DNA was being processed for his own case in the same laboratory facility as the cold case.
The laboratory denied cross contamination and defended their lab practices. They stated quite firmly that the samples had never even been on the same side of the laboratory, and the samples had been carefully contained, and so on. The defense tried to use this to discredit the credibility of the other evidence (the sweat stains), but were unsuccessful.
Regardless, it sure makes one wonder how many times cross contamination does happen.
CSI is pretty freakin’ bogus most of the time, but this was actually based on a real case in Germany. Here’s an article on the incident.
I’ve never worked as a forensic scientist, but I did work with DNA in a lab doing molecular biology experiments for close to 10 years. I don’t watch CSI, so none of my information is coming from there. 
As Chief Pedant mentioned, DNA testing is very accurate with respect to identifying the individual who left behind the DNA sample. (Of course, individual cases are fact-specific, e.g., the relevance or weight of the DNA evidence depends, for example, on whether the DNA was located under the victim’s fingernails or simply present somewhere at the scene.)
Although other methods are sometimes used, the most common forensic DNA analysis method in the US is STR (short tandem repeat) analysis. All DNA profiles stored in CODIS (see below) are generated using STR.
Bolding mine.
However, there can be technical problems with DNA analysis. For example, contamination with other evidence (secondary transfer) is a key source of incorrect DNA profiles and raising doubts as to whether a sample has been adulterated is a favorite defense technique. In addition, the ability of STR to discriminate between two people is lower if the actual source is a person related to the suspect, especially a sibling. While almost all individuals have a single and distinct set of genes, rare individuals, known as “chimeras”, have at least two different sets of genes. There have been several cases of DNA profiling that falsely “proved” that a mother was unrelated to her children.
The location of the DNA evidence does not suggest it was due to incidental contact of the kind you suggest. There was DNA in a drop of blood on JonBenet’s underpants. A separate lab also found matching DNA in two locations on her longjohns (different than her underpants). This DNA was found by scraping the waistband and the sides of the longjohns - the locations where someone would have touched when pulling them off. The amount of DNA identified on the longjohns was enough for it to be processed in the routine way DNA is analyzed. (In other cases, so-called “low copy number DNA” has to be processed in a different way, using more cycles of PCR amplification of the DNA). Cite.
DNA from the two sites on the long johns matched genetic material from an unknown male that had previously been recovered from the blood in JonBenet’s underpants. Additional tests were conducted to ensure that the genetic material did not come from law enforcement or medical examiner’s personnel. Cite.