Learning Molecular Representation in a Cell
This work addresses the challenge of predicting drug efficacy and safety in vivo by improving molecular representation learning for biological responses, offering a novel method with strong specific gains in domain-specific applications.
The authors tackled the problem of learning molecular representations that capture comprehensive cell states under small molecule perturbations, introducing the InfoAlign approach which achieved superior performance in molecular property prediction against up to 27 baseline methods across four datasets and enabled zero-shot molecule-morphology matching.
Predicting drug efficacy and safety in vivo requires information on biological responses (e.g., cell morphology and gene expression) to small molecule perturbations. However, current molecular representation learning methods do not provide a comprehensive view of cell states under these perturbations and struggle to remove noise, hindering model generalization. We introduce the Information Alignment (InfoAlign) approach to learn molecular representations through the information bottleneck method in cells. We integrate molecules and cellular response data as nodes into a context graph, connecting them with weighted edges based on chemical, biological, and computational criteria. For each molecule in a training batch, InfoAlign optimizes the encoder's latent representation with a minimality objective to discard redundant structural information. A sufficiency objective decodes the representation to align with different feature spaces from the molecule's neighborhood in the context graph. We demonstrate that the proposed sufficiency objective for alignment is tighter than existing encoder-based contrastive methods. Empirically, we validate representations from InfoAlign in two downstream applications: molecular property prediction against up to 27 baseline methods across four datasets, plus zero-shot molecule-morphology matching.