Simplifying Molecular Space Exploration with AI and Synthetic Data

As mentioned earlier, exploring the entire molecular space based on physics alone is practically impossible due to its size (Figure~/ref{fig:2222}) /cite{Chuang2022,COLEY2021133,Lipinski2004nature,Kang2018}. In the construction of GIBAC, a hybrid approach combining AI and physics is used. However, AI algorithms still struggle to predict intermolecular binding data that are different from the training data set, which only covers a limited portion of the molecular space /cite{Kang2018}. To address this, synthetic data and its generators, such as Chen et al. /cite{Chen2021Synthetic,Jadon2023}, are utilized.