"The probability of randomly selecting an unrelated individual with this DNA profile is 1 in 9.2 quadrillion." Similar statements are heard daily in courtrooms across the country. To most judges and juries this sounds to be pretty convincing (and possibly damning) evidence. But, what would you say if someone told you that a search of the state DNA database in Arizona turned up not one, not two, but 144 DNA profiles that matched each other at nine or more loci? Would you be surprised? It turns out the real question might be, "Should you be surprised?"
Arizona-Style Searches
It is a fact that a court-ordered search of the Arizona state DNA database revealed that 122 DNA profiles matched each other at nine loci and 20 DNA profiles matched at ten loci. In addition, one match each (both turned out to be between full siblings) was revealed at 11 and 12 loci. The FBI requires that at least 13 loci, or locations, on the DNA molecule be analyzed for entry into CODIS. Many labs go beyond these "core CODIS loci" and look at a total of 15 locations. The search was ordered by a California judge after a San Francisco defense attorney stumbled across a poster presentation by an Arizona lab worker regarding two DNA profiles in the convicted offender database that were found to match at nine loci, yet the two offenders did not appear to be related (one was white, the other African American). The attorney was representing a client who had been implicated in a crime through a database hit alone and felt that this nine-locus match discovered in Arizona was truly startling and could mean that her client was in fact innocent, regardless of the DNA match.
Expected? Or not?
It seems, at first blush, that this is an extremely anomalous situation. After all, what good are the astronomical odds presented in the courtroom if multiple "matches" are found in the database? To understand how this is possible, one must understand two things. First, the "matching" profiles were only partial matches. The profiles did NOT match each other at the remaining tested loci. In forensics, only one locus must be non-matching between a set of DNA profiles in order for the results to be considered a 100% exclusion. Second, it must be understood that searching a database of profiles against each other for matching profiles is a very different thing than finding a DNA profile at a crime scene and then finding the same DNA profile in the database. The Arizona offender database consisted of 65,493 DNA profiles. This means that 2.1 billion pair-wise comparisons were conducted. Each comparison that is made increases the likelihood that a match will be found. In addition, some of those comparisons would have been between related individuals, which automatically would have increased the likelihood of finding similar profiles. Indeed, as it turns out, the number of observed matches is actually expected! Using the appropriate statistical calculations, Steven Myers, a senior DNA analyst with the California Department of Justice, and Dr. Charles Brenner both (separately) arrived at the conclusion that the number of nine locus matching profiles is not an anomaly, but rather is quite within the range of expected matching profiles. Again, it is important to note that while these 122 profiles matched at nine loci, they did NOT match at the remaining four or more loci, and therefore would not have been reported as matches in a real case.
FBI reluctant to release information
The FBI, which maintains control over the CODIS database system, has been reluctant to perform these "Arizona-style" database searches, and has in fact fought (sometimes successfully and sometimes not) numerous court orders requiring various states to perform them. The argument is that the searches take up too much time and manpower and also require states to go "offline" from the CODIS system for a week or more while the searches are being performed. This would hamper law enforcement's ability to perform DNA database searches and potentially solve crimes. In addition, the FBI argues that the results of these types of searches are both irrelevant and misleading.
In most instances these nine- and ten-locus matches uncovered in Arizona (and later in both Maryland and Illinois through defense-initiated court orders) should not necessarily be seen as a problem or concern when prosecuting a case. Typically a DNA profile will consist of 13-15 loci. These additional loci serve to decrease the probability of finding a matching DNA profile. On the other hand, one area for potential concern, based upon these nine-locus matches, is in the case of a partial DNA profile match. This means that the lab was unable to get results at the full 13 or 15 loci examined.
When there might be a problem
As an example, let's say that analysis of a cold case resulted in a profile of only seven loci. As it is a cold case, there are no suspects. A comparison of the forensic profile with the state database (here it should be noted that at least ten loci must be present in order to perform a search of the national DNA database) indicates a match between the crime scene profile and a convicted offender at the seven loci that are available for comparison. In this particular instance, it may indeed be conceivable that the match is due purely to coincidence especially if no circumstantial evidence also implicates the individual. The less loci that are available for comparison, the greater the chance of an adventitious match. As more and more profiles are added to state databases the chance of a coincidental match, especially when dealing with partial profiles, increases. If additional evidence materials are available for testing, this is an example of a case that screams for more testing. It may be possible to obtain a profile from a secondary piece of evidence that either includes or potentially excludes the suspect at the additional loci. In addition, as always, good old-fashioned detective work may help add to the circumstantial evidence in the case which in turn can help back up the DNA evidence.