---

Specific reference is given to requests of the Commission of Inquiry into Forensic DNA Testing in Queensland of STRmix and PowerPlex21 to consider and review validation, best practise and deficiencies in the period spanning their implementation in 2012 until 2022.

Reports served in evidence to the Commission of Inquiry concluded that;

5. Flawed interpretation of data was evident in both validations. [1]

7. Implementing a threshold that was unsupported by the data created a great risk of not detecting potentially probative, exculpatory or otherwise informative profiling results. [1]

220. The negative consequence to overestimation [number of contributors to a DNA profile] is to incorrectly fail to exclude (and sometimes include with low levels of support) non donors of DNA. [2]

The power of DNA to inculpate, is equalled only by its ability to exonerate.

The authority of this forensic evidence was apparent right from the very first DNA conviction, when Colin Pitchfork was identified using DNA, and jailed for double murder in 1988. Fortuitously, the matter also delivered the very first DNA exoneration, when an alternate suspect was found to have falsely confessed to the crime.

There is little doubt that DNA evidence can be utilised to persuade a jury of guilt, even in circumstances where the scientific evidence itself is of weak probative value.

Mixed DNA profiles

Circa 2012, various probabilistic genotyping software suites were implemented, enabling the interpretation of complex DNA profiles that were not previously assessable. When previously conducting binary assessments, we may have considered dozens of combinations, whereas the genotyping software is frequently called upon to consider hundreds of thousands of combinations.

Probabilistic genotyping software currently utilised by region;

United Kingdom – EuroForMix, LiRa and STRmix.

United States of America – TrueAllele and STRmix 

Australia – STRmix

New Zealand –  STRmix

DNA False Positive

When considering the identify of an individual there is inherent risk of obtaining a false positive. This phenomenon, when referring to the assessment of a mixed DNA profile is known as a false inclusion.

In the early 2000’s, we were dealing with mixed profiles containing say 2 or 3 contributors, to a situation where we are now regularly and readily testing and interpreting mixed DNA profiles with 5 or 6 contributors.

The increased number of contributors within the profile, have amongst other things, contributed further to, and created greater risk of a DNA false positive(false inclusion).

Flawed interpretation of data

An individual laboratory sets the threshold within the software, being the minimum amount of DNA required within a sample to consider continuing with, or abandoning assessment.

In determining the threshold, the lab must rely upon validation data.

Incorrectly set, this threshold setting can result in flawed interpretation or a critical failure to appropriately include or exclude.

In the case of QHFSS, the threshold was set at 132pg.

“at 50pg near to full profiles were obtained, and even at 10 pg half the alleles were detected”.[2]

A higher threshold setting will result in reduced costs and decreased turnaround times, given the inherent reduction in the number of profiles assessed.

“ Likely, the interpretation by QHFSS is driven by bias of maintaining the 132pg threshold”. [2]

“ Implementing a threshold that was unsupported by the data created a great risk of not detecting potentially probative, exculpatory or otherwise informative profiling results” [2]

Mixed DNA profiles – Assigning number of contributors.

In order to evaluate a mixed DNA profile, the analyst needs to assign the number of contributors.

This assignment of contributors is a subjective determination, and critically relies upon factors such as the analysts experience, skill and the standard operating procedures set within the particular Laboratory.

Published literature supports that it is not uncommon for the incorrect number of contributors to be assigned.

“The negative consequence to overestimation [number of contributors to a DNA profile] is to incorrectly fail to exclude (and sometimes include with low levels of support) non donors of DNA” [2]

How Good Science Fails

The failures as QHFSS, whilst sensational, have provided in my opinion a useful window into the potential critical issues that can arise, and more importantly how such issues can seemingly go undetected for many years.

Probabilistic genotyping software has enabled assessment where it was previously beyond reach. It has decreased cost and improved turnaround times.

On the other hand, it has carried with it a propensity for governmental organisations to rely upon the results as though they are infallible, which they are not. Furthermore, there is an increasing number of analysts that do not possess skills and experience predating software use. This skills gap can limit an analyst’s ability to make the necessary subjective decisions as to inputs, and similarly impede capacity to identify issues with outputs.

[1] Review of PowerPlex® 21 & STRmix™ v1.05 Validations – 20 November 2022.

[2] Duncan Taylor – QH STRmix use review – 21 November 2022.