Probabilistic genotyping in forensic DNA analysis

Since forensic DNA probabilistic genotyping (PG) software was introduced nearly a decade ago, its use by federal, state, and local forensic laboratories has increased substantially.

PG software makes better use of the information contained in a DNA profile than traditional approaches. As a result, DNA profiles once discarded as uninformative are now accessible for inculpating and exculpating potential donors of forensic biological evidence. Users of PG software have been able to improve their ability to interpret low-level, degraded, and mixture DNA profiles, positively impacting interpretation of DNA results from a wide range of evidence.

Benefits of Enhancing Interpretation of DNA Evidence

The advent of PG software has enabled forensic analysts to interpret more challenging DNA profiles with greater efficacy and at a much higher degree of accuracy than any interpretational approach previously used.

PG software has proven particularly effective in producing usable, interpretable, and valid DNA results in criminal cases, and excluding individuals wrongly associated as the source of crime scene evidence. PG tools have also been instrumental in helping to solve cold cases in which evidence originally dismissed as inconclusive could be reanalyzed, helping to develop investigative leads and support exonerations in post-conviction cases.

Employing methods like Markov Chain Monte Carlo that are routinely used in computational biology, physics, engineering, weather prediction, and the stock market, PG software grades proposed profiles on how closely they resemble or can explain an observed DNA mixture profile. From there, the probability of the observed DNA evidence can be calculated under two propositions – if the DNA originated from a person of interest or if the DNA originated from an unknown donor. These two probabilities are then presented as a likelihood ratio.

PG software has been validated, as described in a large number of peer-reviewed papers, supporting general reliability. These supporting data have been a foundation for the admissibility of PG software in an overwhelming majority of court cases. Issues raised (such as the software being a black box, access to source code, and potential miscodes) have not gained traction because the supporting data demonstrate scientific validity.

Limitations to Consider

The primary limitation of PG software, as with any methodology, is that it cannot perform miracles. Some DNA profiles have too little information or are too complex to interpret with any PG tool. It is incumbent upon users to perform validation studies to understand the limitations of the tools they employ. By doing so, users will reduce chances of interpreting software output that is not supportable. It is important to understand that the trained user is an integral part of the interpretation of DNA evidence.

Another limitation is the false belief that using PG software renders all such interpretations completely objective. The choice of algorithms and programming are by humans and thus, subjectivity is inherent in the software. Using such software, however, can reduce the variation among users and decrease user variation community-wide.

Challenges to Admissibility

Despite the demonstrated power of PG software, it is not unexpected that, as with most methods in the forensic field, some criticisms have been encountered. First and foremost, our legal system promotes criticism of methods used in legal proceedings, which benefits all and should be embraced.

Second, advocates for the use of a method should perform the proper studies, understand the fundamentals of the method, and apply it properly. If such practices are undertaken, a laboratory should have confidence that its use of PG software is reliable.

Third, some criticize PG software, claiming it has not been subjected to adequate independent peer-review because the inventors are authors of most publications. Peer-review is an important part of science. A paper with an inventor as the author is information, and should be considered in context with papers by others and with validation studies undertaken by each laboratory.

It also is important to recognize that publication of a paper is just one part of the peer-review process. All publications (and presentations at scientific meetings) inform the community, foster discussion, and move science forward. The more effective part of peer-review comes once papers are published. If data and interpretations seem suspect or improvements can be identified, the greater scientific community can comment, critique, and perform studies to demonstrate flaws or alternatives. We should not be misdirected that such authored papers do not provide meaningful data, as the initial paper is only part of the peer-review process. To date, the overall peer-review process has favored that PG software can generate reliable results when used properly.

Finally, some assert PG software contains (or may contain) miscodes which may impact its reliability. The only software that can realistically claim to have no miscodes is the software that has not been written. Stating there may be a miscode is likely to be true. The real questions, though, are: 1) have there been and/or are there mechanisms in place to detect miscodes; 2) what is the impact of a miscode when identified; and 3) are identified miscodes rectified? While the source code of some PG software has been reviewed by individuals outside of the commercial entity, miscodes to date have been identified almost exclusively by examining extended output, which contains the intermediate steps of the interpretation process.

Despite some criticisms, the use of PG software is an advancement that substantially improves DNA analysis. It will undoubtedly continue to have a profound impact on criminal and civil investigations, providing reliable data from a broader range of DNA evidence and in particular complex mixtures.

To promote obtaining reliable interpretations, forensic laboratories need to: 1) ensure their scientists receive proper training on the principles, practices, and data generated by PG software; 2) properly validate their PG software, which includes in-house studies and reliance on peer-reviewed literature; 3) implement effective protocols so as not to overstate the strength of PG results; and 4) address criticisms that are raised which have merit.

Bruce Budowle, PhD is a Professor at the University of North Texas Health Science Center, where he is involved in the research and validation of forensic genetics and molecular genetics methodologies. He also is Director of the Center for Human Identification. Prior to his current positions, Dr. Budowle served 26 years with the Federal Bureau of Investigation. Over the years, he held a number of positions within the FBI Laboratory, including research chemist, chief of the Forensic Science Research Unit, and senior scientist in biology.