How Forensic DNA Software Passes the Daubert Standard
This article was originally published in Forensic Magazine.
By Richard Wivell
In a recent Forensic Magazine article, Vernon Geberth summarized the evolution of DNA technology over the past three decades, concluding with the most recent development – forensic software that allows experienced DNA experts to resolve mixed DNA profiles previously thought unresolvable.
Clearly, sophisticated forensic DNA software represents a huge advance for forensic analysts, significantly improving their ability to contribute to criminal investigations in which DNA evidence formerly was considered too complex to interpret. This means that low-level, degraded, or mixed DNA samples can now be analyzed to maximize the value of every sample collected.
As with most new developments, however, concerns have been raised that advanced DNA software is too new to be used in criminal cases. Relying on the Daubert Standard, some prosecutors and defense attorneys have asked whether this software meets the general acceptance test. Has it been validated independently? Has it been subject to peer review?
The answer to these questions is a resounding yes. But before examining these points, let’s take a quick look at how this software works and why its use represents such a breakthrough in criminal investigations.
As noted, sophisticated forensic software was developed to help standardize the interpretation of mixed DNA profiles and extend a DNA analyst’s ability to interpret complex DNA profiles. STRmix, for example, uses a scientific understanding of the behavior of biology to develop millions of conceptual profiles within the software. It then grades the profiles on how closely they resemble the observed mixed DNA profile. Successful conceptual profiles are examined to identify the individual DNA profiles that best describe the observed mixed DNA profile. Each of these individual DNA profiles can then be compared to a suspect’s DNA profile in order to calculate the strength of the evidence when compared to these profiles.
This approach enables users to interpret DNA results faster, compare profiles against a person of interest, calculate a likelihood ratio, use more of the information in a DNA profile, and ultimately, resolve previously unresolvable and highly complex DNA mixtures.
To date, sophisticated forensic software has been used in literally thousands of cases. In a recent Michigan court case, for example, the use of forensic DNA software by an analyst provided important evidence in an armed robbery case by aiding the interpretation of a DNA profile obtained from the inside of a sneaker left behind at the scene of the crime.
Despite this successful track record, some attorneys have claimed that DNA software is too new to meet the general acceptance test in legal cases. While it is true that sophisticated DNA software has been in use for less than a decade, it is based on standard mathematics that have been used for decades. The probability models and Markov Chain Monte Carlo (MCMC) methods used by such software were born in Los Alamos, NM during World War II then brought closer to statistical practicality by the work of Hastings in the 1970s. Widely employed outside of forensic science, MCMC is used in everything from computational biology and weather prediction to physics, engineering, and the stock market. Thus, the only thing “new” about this software is the combination of these two methodologies.
What about peer review and validation? One question repeatedly asked in court hearings has been whether the software has been scrutinized by other professionals in the field and independently validated. The answer to both questions is yes. Numerous scientific papers have been published in peer-reviewed scientific journals. In addition, internal validations have been carried out by all laboratories in current casework, as is required by their accreditation.
Beyond that, the International Society for Forensic Genetics published guidelines for validating software in 2016, while the Scientific Working Group on DNA Analysis Methods (SWGDAM) has produced guidelines for validating probabilistic genotyping tools. The Organization of Scientific Area Committees (DNA Analysis 2 Sub-Committee) is also working on standards for assessing probabilistic genotyping software tools.
Another matter raised from time to time by both prosecutors and the defense is that because the algorithms used in the software are secret, there is no opportunity for the functioning of the software to be properly reviewed and challenged. In truth, making code available varies from product to product. Some software providers have successfully defended applications to disclose source code numerous times. Others such as STRmix make code available under a non-disclosure agreement and conduct four-day training workshops to ensure that users understand how it works and are able to correctly represent the results it generates in court. Moreover, all of the algorithms have been published in the literature and are freely available.
Perhaps the most important consideration here is that regardless of specific policies with respect to secrecy, it is never a good idea to make it possible for people to install and use such powerful software without proper training.
Finally, what about the miscodes occasionally found in forensic software? While the existence of disclosed miscodes understandably leads to questioning on the part of legal counsel, the bottom line here is that no software of any complexity is free of miscodes. Typically, though, the miscodes that have materialized in sophisticated DNA software occur on the edges of normal usage and have, to date, had no substantial effect on the statistic given. There is always some untested circumstance, either some aspect never envisaged or some rare set of circumstances that may result in identification of a miscode. So while a 100% guarantee that the software is error free probably doesn’t exist, it is possible to guarantee that normal usage has been extensively tested.
There are certainly other questions attorneys have raised – and will rightly continue to raise – with respect to sophisticated forensic software. It’s the nature of anything new introduced into the courts. The bottom line, though, is that DNA evidence is one of the most effective tools available in modern law enforcement. And while the future is likely to see even more advances, the increasing acceptance and use of sophisticated forensic DNA software will undoubtedly have a profound impact on the rapid exclusion of non-contributors, improving prosecution rates, increasing case clearance rates, acting as a deterrent, and otherwise playing a key role in systematic crime reduction.
Richard Wivell is Senior Scientist, Forensic Biology at the Institute of Environmental Science and Research Limited. A graduate of the Imperial College, University of London, Wivell previously served as a forensic scientist at the Metropolitan Police Forensic Science Laboratory in London. Over the past 25 years, he has presented evidence in numerous court cases. He has also presented at international conferences in Tokyo, Vilnius, Sydney, Adelaide, Auckland, and Singapore. Over the last18 months has been part of the STRmixTM team delivering four day DNA mixture training workshops in USA and Canada.