Researchers are turning to Artificial Intelligence (AI) to understand the human genome better. By combining novel computing methods with extensive genomic datasets, they are developing new ways to understand how genes are expressed in the body to better support health advances in Australia. A team from The Australian National University (ANU), led by Associate Professor Jiayu Wen, have developed their own AI model capable of accurately and quickly predicting how a gene might be expressed based on its DNA sequence.
This process takes advantage of the NCI Graphics Processing Units for maximum parallel performance, using up to 32 simultaneously to further speed up the process. By conducting this prediction step computationally ahead of any experimental work, researchers can narrow the areas of interest and focus more time-consuming and expensive laboratory work on the most likely candidates they identify.
Associate Professor Jiayu Wen says, ‘Our computational research group employs a variety of approaches to study gene regulation and expression. Artificial Intelligence is developing into a useful tool for scientists to investigate many more genome sequences than ever before and in new ways. This work we are doing flows through to bioinformaticians and health researchers, contributing to speedups and discoveries that will help support better health outcomes.’
Supercomputers such as NCI’s Gadi are crucial for developing these technologies, as they have much higher performance processors and significant memory capacity, making the training of large AI models much quicker and easier. As with other AI technologies around the world, this field has seen a huge growth in capability over recent years. Once only a scientific concept, the AI field now offers tangible and effective health improvements in clinical practice.
