Implicit Geometry and Interaction Embeddings Improve Few-Shot Molecular Property Prediction

TL;DR

This paper introduces novel molecular embeddings that incorporate implicit geometric and interaction information, significantly enhancing few-shot molecular property prediction by leveraging synthetic docking data and multi-task learning.

Contribution

It presents a new embedding method encoding complex molecular features, improving few-shot learning performance across multiple benchmarks.

Findings

Enhanced performance of few-shot models with the new embeddings

Effective use of synthetic docking data for embedding training

Improved results across multiple few-shot learning algorithms

Abstract

Few-shot learning is a promising approach to molecular property prediction as supervised data is often very limited. However, many important molecular properties depend on complex molecular characteristics -- such as the various 3D geometries a molecule may adopt or the types of chemical interactions it can form -- that are not explicitly encoded in the feature space and must be approximated from low amounts of data. Learning these characteristics can be difficult, especially for few-shot learning algorithms that are designed for fast adaptation to new tasks. In this work, we develop molecular embeddings that encode complex molecular characteristics to improve the performance of few-shot molecular property prediction. Our approach leverages large amounts of synthetic data, namely the results of molecular docking calculations, and a multi-task learning paradigm to structure the embedding space. On multiple molecular property prediction benchmarks, training from the embedding space substantially improves Multi-Task, MAML, and Prototypical Network few-shot learning performance. Our code is available at https://github.com/cfifty/IGNITE.