Reviewed by Danielle Ellis, B.Sc.Oct 1 2024
Flinders University experts have applied machine learning to DNA profiling, forecasting promising new advances in critical DNA testing. The journal Genes published the study.
This century, PCR (Polymerase Chain Reaction) DNA profiling has revolutionized high-throughput sampling in a variety of fields, including national security, forensic testing, and medical diagnostics. However, not much has changed since the technology was created in the 1980s.
Even a small improvement in PCR performance could have a huge impact on the hundreds of thousands of forensic and intelligence DNA samples amplified every year – notably when samples are degraded.
Dr. Duncan Taylor, Forensic Science, Flinders University
PhD candidate Ms. Caitlin McDonald of the College of Science and Engineering, who conducted the study, found that artificial intelligence technologies significantly improved the quality of DNA profiling and more effective PCR cycle conditions.
Our system has the potential to overcome challenges that have hindered forensic scientists for decades, especially with trace, inhibited or degraded samples. By intelligently optimizing PCR for a wide variety of sample types, it can dramatically enhance amplification success, delivering more reliable results in even the most complex cases.
Caitlin McDonald, College of Science and Engineering, Flinders University
Caitlin McDonald also recently presented the study at the International Society of Forensic Genetics conference.
Caitlin McDonald said, “Beyond forensics, this system has the capacity to revolutionize other fields that depend on PCR, such as clinical diagnostics and environmental monitoring, by boosting efficiency, reducing errors, and enabling high-throughput analysis across diverse applications.”
PCR is a widely used laboratory technique that is used to replicate or amplify tiny segments of genetic material. It can be used in various applications such as DNA fingerprinting, genetic condition diagnosis, and the detection of viruses or bacteria like COVID-19.
Supported by additional experts from Flinders University's College of Science and Engineering, such as Associate Professor Russell Brinkworth and Professor Adrian Linacre, the study used machine learning to develop new “smart PCR” systems that target faster cycling conditions and large-scale potential alterations to produce results that are more accurate and timely.
PCR is widely employed across numerous sectors and applications, including forensic science, animal research, health, and national security, according to Flinders University Professor Linacre, who specializes in DNA forensic technologies.
“AI and machine learning are so new, yet harnessed correctly, have the possibility to greatly increase the sensitivity of PCR testing,” said Professor Linacre.
He claims that since 1994, forensic testing has been conducting studies on non-coding regions of DNA.
“With further research, these AI-ML methods have potential to improve the quality of DNA evidence used in criminal investigations, and to increase the quality trace DNA samples, enhancing the criminal justice process.”
According to Russell Brinkworth, an Associate Professor of Autonomous Systems, enhancing current procedures will help define AI applications going forward.
Traditionally DNA amplification required all settings to be in place before the process commenced. This did not take into account the many possible differences between samples and conditions. By utilizing advances in machine learning and sensors, we have changed the process of PCR from a one-size-fits-all to a customized and optimized individual experience. Producing higher quality and quantity DNA faster than previously possible.
Russell Brinkworth, Associate Professor, Flinders University
Journal Reference:
McDonald, C., et al. (2024) Developing a Machine-Learning “Smart” PCR Thermocycler, Part 1: Construction of a Theoretical Framework. Genes. doi.org/10.3390/genes15091196.