Monday, September 27, 2010

An overview of some DNA evidence in the murder of Meredith Kercher

Part 23 in the Knox/Sollecito case

Executive summary
The most severe problems with the bra clasp are that three other partial profiles are present, that it was moved by unknown means before it was collected, that it was handled way too much during its collection, and that it might be within LCN range. Massei’s reasoning with respect to the disputed loci is fundamentally flawed. The most severe problems with Meredith’s profile on the knife are lack of blood, the low peak heights, the alleles that dropped in and dropped out, and the fact that an inferior version of low copy number (LCN) DNA analysis was used. Amanda’s DNA found with Meredith’s blood is not inculpatory, because DNA in one’s home is rarely out of the ordinary, and there are collection problems with these samples. Finally, the electronic data files were not released, and this has hampered the ability of the defense to challenge the evidence properly.

General defense arguments
Besides the argument that a profile is insufficient to identify a person (see below), an argument that the defense can use is that the DNA is present due to innocent activity. For example, Amanda’s DNA on the handle of the knife may be due to her using it to cook. In addition, the defense can argue three things with respect to how a defendant’s DNA came to be on an item of evidence: secondary/tertiary transfer, contamination, and evidence tampering.

The legal standard for a DNA profile and low copy number DNA
One argument that the defense can use is to say that a given DNA profile fails to clear the legal bar for identifying someone. One might use the analogy to a partial fingerprint. An example here is the necessity of testing low copy number (LCN) DNA twice, whereas the knife was only tested once. The bra clasp falls into a gray area in terms the amount of Raffaele’s DNA present; the defense is arguing that it, too, falls into the LCN range. Meredith’s profile on the knife also shows other evidence of being in the LCN range, such as peak height imbalance. Moreover, the majority of the peaks are below 50 RFU in peak height, most labs have set a threshold of at least 50 RFU as the minimum height for a peak to be counted. Meredith’s DNA profile should not have been accepted by the court as evidence, and Raffaele’s lawyers may have success with their argument.

If we only allowed the contamination argument in cases where the defense could demonstrate the exact mechanism of when and how it happened, we would exclude known cases where it did happen. The prosecution must perform negative controls under the same conditions as the evidence and disclose the results of these control experiments to the defense. If DNA shows up in the negative controls, my understanding is that all evidentiary samples processed at the same time must be performed over. When no rational explanation for the presence of DNA on an item can be offered, one is forced to assume that the DNA arrived via contamination. Some cases of DNA contamination are the Jaidyn Leskie murder, the Jane Mixer murder, the Farah Jama rape case, the Gregory Turner case, and the profile N case in New Zealand.

The lack of frequent glove changes and the handling of the clasp by many forensic technicians are problematic for the prosecution. Moreover, Dr. Stefanoni’s testimony as presented in the Massei report (pp. 202-203 in the English translation at Perugia Murder FIle) on this subject is open to serious challenge. Her argument is that they did not change gloves with respect to a certain item of evidence and this piece of evidence did not show contamination. Therefore, contamination is difficult, requiring either liquids or vigorous rubbing. This view seems to be at odds with the consensus of the field, as well as the facts of the Gregory Turner case, which involved transfer of DNA from fingernails to a wedding ring.

Confirmation bias
Not doing what amounts to substrate controls on the mixed Knox/Kercher DNA samples might have been due to confirmation bias. Not obtaining DNA reference samples from Laura and Filomena are behaviors consistent with confirmation bias. Using a lower peak height threshold for the knife than any other piece of evidence contradicts the words of an introductory textbook on DNA forensics as well as general scientific principles. However, there is an additional reason to suspect that some form of investigator bias was at work with respect to the knife profile. Sample 164 was blood from the wall of a bedroom, but it was not tested because of a “negative preliminary (quantification) result.” One surmises that there was not enough DNA to continue the forensic analysis. Why should testing have been stopped for item 164, which had blood, when it was continued for the knife, which had no blood?

Expert testimony
The prosecution’s witnesses and the defense’s witnesses do not have equal scientific standing. Dr. Stefanoni has not published any articles on DNA forensics of which I am aware. On the other hand the nine signers of the open letter (Dr. Johnson, Dr. Hampikian, and the seven co-signers) publish regularly. That is not to say that publication record or academic prestige is everything; there is also variation in the abilities of expert witnesses to convince a jury that they are correct. In this matter the defense may have not fared as well.

The bra clasp
The following discussion assumes that a full DNA profile corresponding to Raffaele Sollecito’s is present, but that does not answer the question of how or when it got there. One problem with the bra clasp as evidence against Raffaele is that his DNA is not found on the bra itself, a point that his lawyers raised in the Micheli preliminary hearing. A more serious problem is that partial profiles of three other people have also been reported.

An answer to the question of how Raffaele’s DNA was deposited on the clasp is that it might have been deposited the same way as three DNA profiles from unknown individuals. To put it another way, if we acknowledge that DNA from three unknown individuals came to be on the clasp innocently, then what makes Raffaele’s DNA different? It is very unlikely that four people handed the bra clasp as part of a sexual assault and murder. Primary transfer before the assault seems equally unlikely; most people fold their own laundry, and someone else folding it would only account for one profile. So the unknown DNA had to arrive either from secondary transfer or from contamination.

One possibility involving secondary transfer involves the towels. The towels that Rudy probably took from the bathroom might have had DNA from anyone who washed his or her hands in the bathroom and used the towels to dry off, including Raffaele, who had cooked there. If the towel were placed over the clasp and stepped on, it could transfer DNA to the clasp. This might also explain the deformation of the clasp. Another possible route of secondary transfer is that whoever moved the clasp before it was collected deposited Raffaele’s DNA (possibly originating from the door).

There is a good deal of misunderstanding involving DNA contamination and the clasp. It is often said that the only item that tested positive for Raffaele’s DNA was a cigarette butt, and so how could contamination occur? There are several problems with this argument. First, one should not equate the DNA that the investigators found with the total amount of DNA Raffaele left at the cottage. The investigators seemed to be focused on blood, as well as Meredith’s body. They were not taking a random sampling of the cottage. Second, there is no reason to exclude contamination from the cigarette butt in the lab, although if they were tested far apart in time, contamination is less likely. Third, Raffaele’s reference sample is a potentially serious source of contamination. In PCR-based DNA forensics, the DNA is amplified very roughly a millionfold in amount. That is why good labs separate the pre-PCR from the post PCR-samples.

The amount of Raffaele’s DNA on the clasp is borderline LCN. If it is judged to be below the LCN cutoff, it would ordinarily have to be tested twice and only those alleles that showed up in both runs should be counted. It is sometimes said that the amount of Raffaele’s DNA was so large as to mean that it could only arise from vigorous rubbing (primary transfer). The fact that the amount of DNA is actually low would seem contradict such an argument. However, it is the premise that is wrong; the DNA profile itself can rarely give an indication of when and how it was deposited. One cannot rule out primary transfer when the amount of DNA is low any more than one can rule out secondary transfer when the amount of DNA.

Raffaele’s appeal with respect to the bra clasp
The discussion above presumes that a good, complete profile was found. However, Dr. Tagliabracci disputed that the profile matched Raffaele’s for at least six of the loci. If Sollecito’s profile were strong and if the bra clasp DNA were not a mixture, there might be fewer opportunities for disagreement between Dr. Tagliabracci and Dr. Stefanoni. Of the six disputed loci from the bra clasp DNA profile, Massei wrote (pp. 296-297 of the Perugia Murder File English translation):

“Consequently, there are apparently a considerable number of loci that are not the subject of dispute, a number which seems to be greater than the number of disputed loci and greater than the number of six loci with reference to which Professor Tagliabracci had previously declared, before the current systems were available‚ it was enough ... we made hypotheses even with six loci‛ (page [319] 103). The circumstance now exposed allows, it was held, the following consideration: if, despite the subjective contribution of the geneticist, the interpretative disagreement regarding the non-compatibility of Raffaele Sollecito’s profile with the loci that had contributed to forming trace 165B involved those loci indicated by Professor Tagliabracci during the course of the hearing and at pages 20 and 21 of the previously mentioned memorandum conclusions, it must be held that, for the greatest number of loci at least, the peaks were so clear and the interpretation so sound that they could not be contested. Consequently, the overall result should be considered fully reliable, even disregarding the repetition of the analysis. It should however be noted that Dr. Stefanoni, during the hearing at which she testified, had offered suitable explanations and answers which this Court considers acceptable.”

Raffaele’s appeal document correctly notes that Massei’s argument about the numbers of disputed and undisputed loci is contrary to the principles of forensic genetics. Let us assume that the data are clear enough to avoid ambiguity and consider the following analogy. Suppose that a winning lottery number is 12497635834, and I have a lottery ticket that is 12497235834. And suppose I claim that since my ticket has 10 out of the 11 numbers identical, I am a winner. That argument makes as much sense as Massei’s does.

But what of Dr. Tagliabracci’s statement that six loci used to be enough to form hypotheses? Suppose that initially a complete profile consisted of six loci. If a person matched all six loci, he or she would not be excluded as the DNA donor to that sample. However, if that person matched at only 5 loci and failed to match at the sixth locus, then he or she would be excluded. Now suppose an improved test with 10 loci became available. Then a person who matched all ten loci would not be excluded, and the number of other people who could also match would be much smaller than in the case with 6 loci. However, a person who matched at 9 loci but failed to match the tenth locus would still be excluded, even though 9 is greater than 6.

Massei must believe that at all six disputed loci, the DNA is Raffaele’s, or at the very least that the results in all six loci are indeterminate (if the latter were true, it would indicate that Raffaele’s DNA constituted a partial profile, not a complete one). Massei does not provide a clear reason for rejecting Dr. Tagliabracci’s assessment in favor of Dr. Stefanoni’s. It is difficult to see why a sentencing report the fails to provide reasons is any better than no sentencing report at all.

The knife profile
The peak heights on the DNA profile culled from the kitchen knife are all below 100 relative fluorescence units (RFU), and most are below 40 RFU. This is below any threshold of which I am aware. What was the harm in using a lower peak threshold? One can argue that it obliges the forensic scientist to use the same threshold for all the samples on the basis of consistency. It is a dollars-to-donuts bet that some evidence of contamination could be found at this atypically low peak threshold among the hundreds of samples run.

There is no detectable blood on the knife. The open letter asserts that if a bloody knife were cleaned, one would remove detectable traces of DNA before detectable traces of blood. If one claims that the DNA arose from other tissue, then I would ask how it is possible to remove blood cells and not other cells. The cleaning problem only grows more severe if one claims that the knife were cleaned with bleach and that traces of bleach were found. Even trace amounts of bleach are known to destroy DNA for forensic profiling.

The profile shows evidence of alleles dropping in and dropping out. In other words there is one allele where Meredith’s profile is weak or absent, and there is one locus with two peaks that are not part of Meredith’s profile. The peaks within each locus are often very uneven (as much as roughly threefold) in peak height, yet they should be approximately the same height in a good profile. These problems are to be expected when DNA is in the low copy number (LCN) range. When DNA falls into such a low range of amounts, forensic scientists generally test it at least twice and accept only those peaks that appear in both runs.

One can argue that LCN profiling should ordinarily be accepted by a court. However, Dr. Stefanoni used an inferior version of LCN DNA profiling, one that has never appeared in the scientific literature. LCN profiling is typically done in specialized buildings, away from the laboratory doing regular profiling. These precautions are necessary because LCN profiling is more prone to contamination than ordinary PCR-based profiling. These precautions were not followed with respect to the knife, and it was only tested once.

Another problem with the knife is that the second officer to have possession of the knife was at Meredith’s cottage just before receiving it. This raises the odds of contamination outside of the lab. Meredith’s profile probably arose through contamination in the laboratory, but contamination during the time it was taken into custody is also a possibility.

The mixed DNA samples
A number of samples that appeared to be blood had both Meredith’s and Amanda’s DNA. Three of the mixed DNA samples were probably blood and three may or may not have been blood. This would be very weak evidence under most conditions. Amanda’s DNA is expected to be in many locations in her own home. The fact that samples were not taken close to the blood (essentially substrate controls) means that one cannot rule out an innocent explanation for their existence. To argue that these samples are inculpatory, one is almost forced to assume that Amanda’s DNA is from her blood. White blood cells contain DNA; therefore, Amanda’s DNA might have arisen from her blood. Yet without characterizing or quantifying the amount of biological material that gave rise to her DNA, there is no reason to believe that the samples must be from blood. No such tests were done.

However, the prosecution has two additional problems with its case. First, Dr. Stefanoni did not change gloves when collecting multiple samples (see above). Therefore, she might have mixed samples herself. Second, at least one of these samples had a third profile in it, from an unknown individual. If this person’s DNA arose from innocent means, there is no reason to exclude the possibility that Amanda’s did also.

The lack of DNA
The lack of Raffaele’s DNA or Amanda’s DNA on Meredith’s body, when Raffaele is thought to have restrained her and Amanda to have throttled her calls into question this part of the prosecution’s narrative. The number of actual instances where DNA was used in strangulation cases where the DNA originated from the victim’s neck, as opposed to the victim’s fingernails, is small. However, some instances of alleged domestic violence cases have used swabbing of bruised or reddened areas on the alleged victim as evidence.

The way that the bra clasp was handled and the lack of a clear chain of custody cast doubt on this piece of evidence. The lack of blood on the knife calls into grave question whether the DNA got there before or after the police took it into evidence, as argued in the Johnson/Hampikian open letter. Secondary transfer is a likely means for Raffaele’s DNA being in the bra clasp, and contamination, either in the lab or during collection, is a likely means for Meredith’s DNA being found on the knife. The mixed DNA samples are virtually meaningless. The single most troubling aspect of the DNA evidence is the lack of full disclosure of the electronic data files and other documentation relating to the DNA forensics. The prosecution is acting as if it had something to hide.