DNA Databanking: The Next Step

Catalog | Cart | Log In

DNA Databanking: The Next Step

Paul B. Ferrara, Ph.D., Director
Virginia Division of Forensic Science

× Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø

The forensic science community now stands poised to fully assume its proper role as a major investigative tool for law enforcement agencies throughout the country. The limited successes of RFLP based databases composed of only a small segment of convicted felons has provided a useful insight into the full potential of this technology yet to be realized. An even more tangible view of the future of DNA Databanks in the U.S. is available to us today by looking at the success of the databank program in the U.K. To take maximum advantage of the technologies available to us today, we must take several important steps:

Expansion of the existing state statutes to include all convicted felons and juveniles. In Virginia, we have already seen that more than one-half of all of our cold hits against our convicted felon database have identified rapists and murderers whose sample was in our database as a result of a prior conviction for burglary or breaking and entering. Juvenile offenders, representing a very small percentage of the population, are responsible for a disproportionate number of felony offenses. The Virginia legislature first passed a law in 1989 requiring only certain sex and violent offenders to provide samples for inclusion in a DNA Databank, but a year later, expanded it to all felons including those already incarcerated on the effective day of that law, July 1, 1990. Six years later, the law was further expanded to include juveniles over the age of 14 who were found guilty of any crime which, if committed by an adult, would constitute a felony.

Over the past seven years, the Virginia Division of Forensic Science has received and stored almost 150,000 blood samples and we continue to receive 20,000 annually. By practice these samples are taken during the intake processing of a felon, and because Virginia is a no-parole state, felons serve an average of six years before they are released and free to commit further offenses. Thus this backlog has not been as problematic as it may appear. Submission of multiple samples from one individual or no samples from others are logistical issues which can be resolved more easily than the necessary determination in other states as to who should be sampled depending upon the nature of their conviction. In Virginia, it’s simple: a felony conviction and you’re in. Our net is arguably larger than necessary, but at least we aren’t missing those repeat offenders who are pleading out to lesser charges not included in more narrow nets.

It was clear to us in 1990 and to the first NRC Committee on DNA Technology in Forensic Science that RFLP technology was not amenable to establishment of large databases. It was also clear to us then that STRs would someday be the foundation to support creation of large databases. That day has arrived. One of the most important provisions of Virginia law has been the ability to retain collected samples even after analysis. The Division has already begun re-analysis of the over 10,000 samples currently profiled by RFLP using STR multiplexes while casework examiners train in this same method. Beginning in July of 1998, the Division plans to out-source some 40-50 thousand convicted felon samples per year in addition to running 10-20 thousand per year in-house. I have submitted a budget proposal for over $10MM over the ‘98-2002 quadrennium which is supported by the current administration as well as both gubernatorial candidates running for election this year. Within four short years, we will have a database in excess of 240,000 convicted felon profiles and we will be able to then maintain currency with future samples as they are received. An additional advantage to this approach will be a contemporaneous transition in the method of collection of samples to the use of buccal or saliva swabs. The cost savings alone in the collection of samples by Virginia’s Correctional personnel are great; but even more significant will be the enhanced safety of corrections personnel, health-care workers and particularly our own laboratory staff who must currently aliquot liquid blood samples from a high-risk population onto stain cards.

Independent laboratories will provide a very important role in the development of large databanks such as that envisioned in Virginia. Despite our encouragement for the defense to seek independent re-analysis of our casework samples, as recommended by NRC II, the fact of the matter is that the defense simply resists this for obvious reasons. Thus, absent passage of a statutory provision for independent re-analysis - unsuccessful thus far in Virginia - I no longer see a significant volume of this type of work for independent laboratories. However, these laboratories do represent a great resource for the development of these large convicted felon databanks. This approach will also allow our examiner staff to concentrate their unique skills and training on the analysis of crime scene samples. One word of caution, however; Virginia, for one, will not contract to a private laboratory who cannot demonstrate compliance with the Quality Assurance Standards for Convicted Offender DNA Databasing Laboratories suggested by the CODIS subcommittee of TWGDAM for promulgation by the FBI’s DNA Advisory Board. These draft standards will be discussed at the September 23, 1997 DAB meeting. I encourage interested parties to attend that public meeting.

Despite the success of RFLP databanks in identifying perpetrators of violent crimes over the past 5 years, it has also pointed out the much greater proportion of cases in which we encounter insufficient amounts of undegraded DNA suitable for RFLP analysis. Time after time we are capable of amplifying DNA and developing a profile of the contributor of genetic material, but yet have not, until now, had a corresponding database to search. Therefore, the real power of STR multi-plexes will be felt not only because of the larger database possible but because of the sensitivity and specificity on case work samples. These advantages will allow us to use STR loci and maintain a constant methodology for a long period of time. Our valuable examiners need to be able to work cases and not be constantly validating and maintaining multiple DNA systems. We will be able to help investigate old, cold cases for which the combination of STR technology and corresponding large databanks will lend closure to countless of these cases; that is, of course, if our criminal justice system has not destroyed the physical evidence from these cases. We should actively encourage proper methods of storage and preservation of at least portions or cuttings of evidence in unsolved cases.

Let me turn now to the issue of reporting the results of matches using DNA Databanks. Clearly, the prosecution of these databank cases center on the DNA databank search and the subsequent direct comparison. Typically, there has been little other direct evidence in these cases, so predictably the focus of the defense has been the DNA evidence and specifically Recommendation 5.1 of NRC II which states, "When the suspect is found by a search of DNA database, the random match probability should be multiplied by N, the number of persons in the database. As a consultant to the NRC II committee and an otherwise enthusiastic proponent of the NRC II Report, I took issue with this recommendation in the final draft. My concerns were realized during prosecution of our 14th cold hit in late 1996 when our database consisted of almost 10,000 convicted felon profiles. Following NRC II recommendations, the suspect profile, with a normally calculated random-match probability of 1 in 3,117,000 in the black population, was reported as 1 in 3,900 in the initial report, despite its uniqueness in a database of almost 10,000 individuals. Predictably the defense filed a motion to dismiss which contained the following points:

1. The rape victim in this case does not and cannot identify the defendant as her assailant.

2. The exclusive evidence allegedly linking the defendant to these offenses is ‘cold hit’ DNA evidence. There is no other corroboration.

3. The Commonwealth’s DNA expert has supplied a written report which suggests these DNA characteristics would be found in 1 of every 3,900 blacks.

4. These statistics are insufficient as a matter of law to exclude every reasonable theory of this defendant’s innocence.

The motion went on to argue that the introduction of the subsequent report of the direct comparison of the crime scene evidence with new sample of the suspect’s blood "would be inconsistent with the requirements of the NRC report and unduly prejudicial to the jury".

Despite the prosecution’s attempts to argue for allowing introduction of the random-match probability contained in the direct comparison report, the judge would only consider one number, that being 1 in 3,900, a statistic four orders of magnitude less than the RMP. The case was dismissed and the defendant released of these charges and the Commonwealth has no further recourse.

The discussion in NRC II following this recommendation states "if the database searched includes a large proportion of the population, the analysis must take this into account". Taking a liberal interpretation of this statement, the Virginia Division of Forensic Science has adopted the following policy on databank hits: When a search of the databank results in a single match, the report will simply state that the search of a databank of X number of profiles resulted in a match. When, based on this probable cause, a direct comparison is made, that final report will be reported using standard random-match probability calculations.

Finally, a few non-expert words on the issue of uniqueness. NRC II admirably opens the door to this issue with the statement that "with an increasing number of loci available for forensic analysis, we are approaching the time when each person’s profile will be unique...neglecting population structure and close relatives, 10 loci with a geometric mean heterogyosity of 95% give a probability greater than about 0.999 that no two unrelated persons in the world have the same profile".

To punctuate this point, let me describe a pending high profile case involving a well-known sportscaster whose case comes to trial next week. Without providing any information that is not already a matter of public record, I can tell you that we first filed the results of comparison of evidential biological fluids to the suspect using DQA1, Polymarker, D1S80 and CTT, deriving a profile calculated a 1 in 2.6 billion. The prosecutor, not wishing to leave any stone unturned, then asked for RFLP analysis which by itself developed a profile calculated at 1 in 8.8 million, which in itself is an interesting insight into the relative specificity of 10 PCR v. 4 VNTR loci. Nonetheless, when these PCR and RFLP results are combined, they lead to a number in the range of 1 in 23 quadrillion, an absolutely incomprehensible number which makes any attempt to suggest anything other than uniqueness a futile exercise. The statistical power afforded us with STR multiplexes, combined with the proliferation and electronic consolidation of large state databases via CODIS should soon lead us to ultimately declaring a profile as unique, thus finally dispensing with confusing and misleading statistical representation of the significance of a match and resultant incomprehensible numbers leading to protracted court hearings.

I look forward to the day when, with the implementation of STR testing, our analysts will be able to examine crime scene material in un-sub cases immediately upon submission, and, after rapidly developing a profile, search the database and report a match within 24-48 hours. Until then, I fear the day when we will find, due to the backlogs in casework and in convicted felon samples, that while those samples were languishing in our backlog, a rapist or a killer, who will be eventually identified via the databank, will also be found to have raped and killed again and again, before we got around to the right samples. If our databanks are to truly be a powerful investigative tool, then they must, by definition, be a rapid tool. The importance of the 24-48 hour period after commission of a crime in solving the crime cannot be understated.

Finally, while I must be dreaming, let me include that our DNA results will be accepted readily by the courts, recognizing that DNA testing is the most objective and reliable form of forensic analysis available today.

Go to proceedings home page