AI for Genomics

Overview: The workshop focused on AI’s role in personalized healthcare, genomic data infrastructure, and solving complex diseases. Along with AI models' technical challenges, the discussion also highlighted challenges in data integration, the need for a genomics-aware workforce, ethical considerations and crucially: how to enhance collaboration between AI and biology experts. Two Singapore initiatives were presented: The TRUST platform for health data storage and exchange, and the PRECISE entity for precision health research.

Vision: The future of AI for Genomics should be an all encompassing AI model that can integrate both multi-modal data and different “scale” of data: from genetic and molecular to medical reports. Furthermore, the model could be extended beyond human studies and DNA sequencing to include plant and microbiome, as well as RNA expression, respectively.

Pivotal in realising the grand vision of AI for Genomics, the workshop highlighted the need for a systematic, standard and scalable means to record, store and share data, whilst ensuring the respect of individuals privacy. The necessity of developing AI models with results that meet existing benchmarks, as well as developing new benchmarks to assess their reliability and ensure confidence in these results from existing practitioners and patients was discussed. Throughout the workshop, specialised LLMs and Deep Learning models were identified as having the most potential for a functional, robust and reliable AI for genomics.