Much of chemistry comes down to a single question: given a molecule, what will it do? Predicting properties and spectra directly from structure could replace slow measurements and calculations, but molecules are awkward objects for a model. They are naturally graphs, yet their behavior depends on the full three-dimensional geometry, which transforms under rotations and reflections.
NMR spectroscopy is a good example of why this matters. Reading a molecular structure out of an NMR spectrum is slow, expert work, and carbohydrates are among the hardest cases because of their structural diversity. We built GeqShift, an E(3) equivariant graph neural network that predicts carbohydrate NMR chemical shifts directly from 3D structure (Bånkestad et al., 2024). By respecting the geometry, it reduces the prediction error up to threefold compared with models that only see the flat 2D structure, and it stays robust even when training data is scarce.
A second difficulty is that labelled molecular data is often scarce, which makes it worth borrowing information from related tasks. In earlier work we pre-trained a transformer on chemical reaction data and showed that this representation improves downstream property prediction across several MoleculeNet benchmarks (Broberg et al., 2022).
Carbohydrates, vital components of biological systems, are well-known for their structural diversity. Nuclear Magnetic Resonance (NMR) spectroscopy plays a crucial role in understanding their intricate molecular arrangements and is essential in assessing and verifying the molecular structure of organic molecules. An important part of this process is to predict the NMR chemical shift from the molecular structure. This work introduces a novel approach that leverages E(3) equivariant graph neural networks to predict carbohydrate NMR spectra. Notably, our model achieves a substantial reduction in mean absolute error, up to threefold, compared to traditional models that rely solely on two-dimensional molecular structure. Even with limited data, the model excels, highlighting its robustness and generalization capabilities.
@article{bankestad2024carbohydrate,title={Carbohydrate NMR chemical shift prediction by GeqShift employing E(3) equivariant graph neural networks},author={B{\aa}nkestad, Maria and Dorst, Keven M. and Widmalm, G{\"o}ran and R{\"o}nnols, Jerk},journal={RSC Advances},volume={14},pages={26585--26595},year={2024},doi={10.1039/D4RA03428G},}
2022
ICML-W
Pre-training Transformers for Molecular Property Prediction Using Reaction Prediction
Molecular property prediction is essential in chemistry, especially for drug discovery applications. However, available molecular property data is often limited, encouraging the transfer of information from related data. Transfer learning has had a tremendous impact in fields like computer vision and natural language processing signaling for its potential in molecular property prediction. We present a pre-training procedure for molecular representation learning using reaction data and use it to pre-train a SMILES transformer. We fine-tune and evaluate the pre-trained model on 12 molecular property prediction tasks from MoleculeNet within physical chemistry, biophysics, and physiology and show a statistically significant positive effect on 5 of the 12 tasks compared to a non-pre-trained baseline model.
@inproceedings{broberg2022pretraining,title={Pre-training Transformers for Molecular Property Prediction Using Reaction Prediction},author={Broberg, Johan and B{\aa}nkestad, Maria and Ylip{\"a}{\"a}, Erik},booktitle={ICML 2022 2nd AI for Science Workshop},year={2022},}
