As a concrete example of drug discovery, we are focusing on antibiotic discovery / design. Discovery and development of antibiotics have significantly reduced the burdens of infectious disease in human population in the past. However, due to natural evolution of microbes and inappropriate usages of antibiotics in healthcare and agriculture, antibiotic resistance has become an urgent problem world wide. In addition, due to various reasons1, development of new antibiotics, especially those with novel structures and targets, has slowed down in the past decades. In the near future, emergence of widely spread antibiotic resistant micro-organism will cause more problems. Therefore, novel methods for antibiotics discovery are urgently needed.
- MUDiff: Unified Diffusion for Complete Molecule Generation, arXiv 2023
- GIMLET: A Unified Graph-Text Model for Instruction-Based Molecule Zero-Shot Learning, bioRxiv 2023
- Med-UniC: Unifying Cross-Lingual Medical Vision-Language Pre-Training by Diminishing Bias, arXiv 2023
- Unifying Likelihood-free Inference with Black-box Sequence Design and Beyond, ICLR 2022
- Bidirectional Learning for Offline Infinite-width Model-based Optimization, NeurIPS 2022
- Biological Sequence Design with GFlowNets, ICML 2022
- Running ahead of evolution - AI based simulation for predicting future high-risk SARS-CoV-2 variants
I’ve become relatively familiar with algorithms for biological sequence design. Next, I will get wet-lab experiments involved.
E.g. most pharmaceutical companies have abandoned the development of new antibiotics and have instead focused on developing more profitable drugs for non-communicable diseases. ↩︎