Lab Data Suppression: Methodology, Sources and Summary

Documents disclosed in the midst of the trial reveal that the prosecution engaged in a massive suppression of DNA evidence relevant to the case. Indeed, the prosecution suppressed in excess of 100 out of 173 genetic profiles that the laboratory processed in just the first month of the investigation. The suppressed results are for batch controls and evidentiary samples, including rape kit evidence and evidence of an additional crime scene in the downstairs apartment. Most critically, there has been a suppression of almost all evidence relating to the time period and batch in which the controversial Rep. 36b (allegedly the murder weapon) was processed, with the available data (and missing data) strongly suggesting that there was a serious problem in the processing of this trace. The prosecution’s suppression of this data deprived the defendants of a full and fair opportunity to investigate laboratory contamination and to examine potentially exculpatory genetic evidence. The concealed records prove a pattern of contamination and improper laboratory practice that eviscerate the prosecution’s key evidence against the defendants.

The analysis shows:

  • The prosecution hid the results of early and decisive DNA tests excluding Raffaele Sollecito as the sexual assailant, securing on improper grounds the pre-trial incarceration of Sollecito and Knox (and Lumumba) to the severe prejudice of the defense;
  • The prosecution concealed the initial results for tests performed on the two key items of evidence, i.e., the kitchen knife (Rep. 36b) and the bra clasp (Rep. 165b), and instead, produced only the results of suspicious “do over” tests (re-runs), without disclosing the data from the initial tests or even the fact that the subsequent tests were “do overs”;
  • The prosecution concealed that the kitchen knife profile was generated within a set of tests for which 90% of the results have been suppressed, strongly suggesting the occurrence of a severe contamination event that the prosecution continues to hide;
  • The prosecution claimed that contamination of the bra clasp was impossible, even though the bra clasp profile was produced during a set of tests for which there is documented proof of contamination;
  • The prosecution falsely portrayed the DNA laboratory as pristine and perfectly maintained, even though the lab’s own documents demonstrate that it was plagued with repeated contamination events and machine malfunctions that were known to the lab;
  • The prosecution has withheld the results from a massive number of DNA tests (well over 100), including probably exculpatory profiles relating to the sexual assault and a secondary crime scene downstairs;
  • The prosecution has hidden all of the records of the DNA amplification process—the most likely place for laboratory contamination to have occurred—including all of the contamination-control tests for this process.

Batch Chart Key


1. The Incomplete Lab Report and Electropherogram Disclosure

On June 3, 2008 Patrizia Stefanoni produced the RTIGF Report of her “technical analysis” (i.e., laboratory work). On September 25, 2008, she separately provided to the court, in the form of a Bates-stamped collection of printed electropherograms, what she later alleged were “all of the electropherograms relating to the genetic profiles extrapolated from the technical analysis.”

Critically, Stefanoni’s electropherograms contain two sets of numbers generated during the profiling process. First, each electropherogram bears a serial “ID” Number, which identifies the order in which the sample underwent amplification.

Colored e-gram of knife traces A and B Page 2

Second, each electropherogram bears a “Plate” Number, which identifies the typically 16- or 24- sample plate in which the sample was submitted to capillary electrophoresis following amplification.

Because the Plate Numbers are sequential, and the ID Numbers are sequential within each plate and from plate-to-plate, it is possible to organize the samples in the order in which they were subjected to profiling, and then to identify “gaps” in both the ID Numbers and the Plate Numbers for the produced electropherograms. The gaps indicate additional amplifications (the “Gap Profiles”) that were not disclosed to the courts or the defendants. However, because Stefanoni in her Report omits reference to the Gap Profiles, and there was only a limited disclosure of information from the profiling process, it was not feasible prior to trial to determine that the Gap Profiles related to the Kercher case or to identify the samples that corresponded to the Gap Profiles.

2. Identifying the “Gap Profiles”

At the July 14, 2009 hearing—more than a year after initially producing her Report and in the midst of trial—Stefanoni was ordered by Judge Massei to produce the additional lab data by July 30, including the results of her quantifications, which can be used to demonstrate that the Gap Profiles relate to the Kercher case, and that permit identification of the individual samples to which the Gap Profiles correspond.

The process for identifying the Gap Profiles starts with the results of the quantification process. The extracted samples were quantified prior to profiling. “Successful” quantifications returned a cycle threshold (Ct) value of less than 37 (in two reactions). Samples that failed to achieve a Ct less than 37 were not submitted to the profiling process. It should be noted that a Ct of 37 represents about one cell of DNA. Thus, as a practical matter, what this means is that the laboratory, as a matter of course, conducted Low Copy Number (LCN) profiling, despite having no validated process to do so. Moreover, the Qubit Flourometer was not sensitive to LCN levels, and all of the Qubit Flourometer samples (e.g., 36b), were simply passed on for further analysis, despite the fact that the lab did not know whether these the samples actually contained any DNA.

Once the “successful” quantifications are identified, the process for identifying the Gap Profiles is straightforward. All of the successfully-quantified samples were submitted to profiling, and during the amplification process, were assigned an ID Number in the same order in which they were quantified. Therefore, a list of all of the successfully-quantified samples, arranged in the order in which they were quantified, serves as an index to the ID Numbers. This index of ID Numbers discloses the identity of the Gap Profiles that are not mentioned in the Report and were unknown to the defendants.

Of course, the actual amplification records and electropherograms that relate to the Gap Profiles were never disclosed, and so they cannot be examined. The indexing process described above, though, shows that the Gap Profiles consist generally of a mixture of negative control samples (“Controls”) and evidentiary samples, many of which are from Low Copy Number (LCN) samples. While the results of these Gap Profiles remain unknown, we can assume that they are not helpful to the prosecution, at best reflecting mixed/uninterpretable results or unexpected/unknown identifications, and at worst showing laboratory contamination or even exculpatory evidentiary results.

3. Batch Charts

Charts are attached below indexing the profiles, including the Gap Profiles, according to the batches in which they were processed.

Batch One (5-6 November 2007)

  • There are three major plates, Nos. 354, 355, and 357, which contain a total of 48 profiles, 14 of which are suppressed, for a suppression rate of 29%;
  • The missing profiles correspond to: 5 traces attributed to “cat’s blood” (which nevertheless quantified positive for human DNA and were subjected to STR amplification), 2 traces representing the male fraction of the semen samples, 2 traces from the lightswitch downstairs, 2 negative controls, 1 sample of Guede’s feces, and 2 traces for which there is no record;
  • Samples of Guede’s DNA were detected in the vaginal swab and feces samples on 6 November 2007. At this time, Sollecito and Lumumba were ruled out as suspects in the sexual assault;
  • All of the samples in plate 357 were quantified via Qubit Flourometer (at the very end of the Batch). The ID and Plate numbers indicate that the corresponding plate was pushed ahead of other plates in the cue for capillary electrophoresis, indicating extreme urgency to process the included reference samples for Knox, Sollecito and Lumumba;
  • Rep. 14 (Sample hair formations taken from the external part of the vagina of the victim) and Rep. 29 (Lumumba Reference Swab) are missing and not in the SAL.

Batch 2 (13-14 November 2007)

  • There are three major plates, Nos. 365bis, 366 and 367, which contain a total of 64 profiles, 57 are suppressed, for suppression rate of 89%. This suggests severe problems with this batch;
  • One of the few profiles disclosed for this batch is 36b, which is one of 4 produced profiles for the 24-sample Plate 365bis (an 85% suppression rate for 365bis profiles);
  • Since “bis” means “redo,” the designation of 36b’s plate as “365bis” demonstrates that this plate was rerun, likely as a result of serious problems in processing or contamination;
  • There are large gaps in the records directly before and after the testing of Rep. 36;
  • All of the samples in Batch 2 are said to be quantified via Qubit Flourometer, although there is no explanation for a gap in RT-qPCR records that would correspond to this time period;
  • Based on its Rep. No. (58), Rudy Guede’s toothbrush was analyzed at the end of this batch, on 14 November 2007, although there are no produced records for the testing;
  • There are no evident negative controls processed in this batch.

Batch 3 (23-29 November 2007)

  • There are three major plates, Nos. 375, 376 and 377, which contain a total of 64 profiles, 36 of which are suppressed, for a suppression rate of 56%;
  • The electropherogram corresponding with ID No. 950 is of note, as it shows a sample of only 35-picograms, but nevertheless, gives a stronger profile than 36b (suggesting that 36b contained less than 35 picograms);
  • The suppressed profiles include two negative controls and e-gram for Rep. 59b – Rudy Guede’s DNA on the bra;
  • Kercher’s mobile telephones would belong at the end of this batch, but no corresponding testing records have been produced;
  • A large number of the suppressed profiles are from the testing of Sollecito’s Audi and luminol hits in Sollecito’s flat;
  • There are numerous unexplained gaps in the laboratory records for this batch.

Batch 4 (13-18 December 2007)

  • Batch 4 contains 72 amplifications, 23 of which are missing, for a suppression rate of 32%;
  • The missing profiles consist of: negative controls (3); items that appear to have been collected from Rudy’s Guede’s apartment (14 samples), samples from Sollecito’s shoes (4); 2 samples of Kercher’s blood, and, curiously, 1 sample of presumed “saliva” from the kitchen knife (Rep. 36);
  • Critically, one of the Negative Controls for this batch shows contamination: see No Template Control located at position B10-12 of RT-qPCR Run No. 564;
  • There are two sets of field-numbering (A to E and 1 to 12) for the samples collected from Guede’s apartment, attesting to the two separate searches of Guede’s apartment (on November 16- and November 20). The police seized a shoe box (presumably for Nike Outbreaks), a towel, toothbrush, sink strainer, jeans and a museum ticket (Items A through E). Curiously, none of the extraction or quantification records for the initial testing of any of the Guede items have been produced, and it is only from the Rep. Nos. (i.e., 58 and 148-163) and a Plate No. (i.e., 403) that we can estimate that the toothbrush, Rep. No. 58, was first tested around November 16, and the rest of the Guede items were analyzed on or shortly before December 13, 2007.

Batch Five (29 December 2007 – January 2008)

Batch 5, which includes the bra clasp (Rep. 165), shows very serious problems. Demonstrable machine malfunction and obvious contamination render unreliable all of the quantification results for Batch 5; in other words, there is no reliable quantification result to show “abundant” DNA in Rep. 165b, as has been claimed. In addition, Rep. 165b (bra clasp hooks) apparently was profiled twice, with the results of the first analysis being suppressed by the prosecution. Record-keeping anomalies, together with the belated re-run of 165b, even suggest tampering with the results of the analyses for this exhibit. Generally, the contamination, machine malfunction, processing irregularities and record-keeping issues render unreliable the results reported for Batch 5, and raise serious questions about the integrity of the lab’s work.


Case Files

Selected DNA results
Polizia Scientifica RTIGF Report
Test results index translated to English
Stato Avanzamento Lavori (SAL)
DNA profiles for Guede, Kercher, Sollecito, and Knox
DNA evidence/test results and e-grams for knife