Highly Accurate Quantum Chemical Property Prediction with Uni-Mol+

Recent developments in deep learning have made remarkable progress in speeding up the prediction of quantum chemical (QC) properties by removing the need for expensive electronic structure calculations like density functional theory. However, previous methods that relied on 1D SMILES sequences or 2D molecular graphs failed to achieve high accuracy as QC properties are primarily dependent on the 3D equilibrium conformations optimized by electronic structure methods. In this paper, we propose a novel approach called Uni-Mol+ to tackle this challenge. Firstly, given a 2D molecular graph, Uni-Mol+ generates an initial 3D conformation from inexpensive methods such as RDKit. Then, the initial conformation is iteratively optimized to its equilibrium conformation, and the optimized conformation is further used to predict the QC properties. All these steps are automatically learned using Transformer models. We observed the quality of the optimized conformation is crucial for QC property prediction performance. To effectively optimize conformation, we introduce a two-track Transformer model backbone in Uni-Mol+ and train it together with the QC property prediction task. We also design a novel training approach called linear trajectory injection to ensure proper supervision for the Uni-Mol+ learning process. Our extensive benchmarking results demonstrate that the proposed Uni-Mol+ significantly improves the accuracy of QC property prediction. We have made the code and model publicly available at \url{https://github.com/dptech-corp/Uni-Mol}.