It's All Connected: Topology-Aware Structural Graph Encoding Improves Performance on Polymer Prediction

2026-05-11

AI Summary

Accurate prediction of polymer properties—such as glass transition temperature—is hindered by the scarcity of labeled data and the neglect of chain-level topological information. To address this, this work proposes a novel approach that samples representative polymer chains according to the Schulz–Zimm distribution, explicitly constructs large-scale graph structures incorporating chain-level topology, and integrates chemical features with PSMILES-based masked graph self-supervised pretraining. Employing GINE and GATv2 architectures for few-shot learning, the method achieves an RMSE of 24.76 K on a dataset of 381 polymers, representing a statistically significant 5.1% improvement (p < 0.001) over pretraining baselines based solely on repeat units. These results demonstrate both the efficacy of the proposed strategy and the architectural generality of the approach.

Abstract

Graph Neural Networks (GNNs) have achieved strong results in molecular property prediction, but polymers present distinct challenges: labeled datasets are scarce and small (typically in the order of hundreds of polymers) due to the need for expensive experimentation, and complex polymer chain distributions influence polymer properties. Established practice in polymer prediction represents polymers solely by graphs of their repeat units, discarding the chain-scale morphology that governs key properties such as the glass transition temperature ($T_g$). In this work, we propose a principled graph construction that addresses this gap. Given a polymer's molecular mass distribution (MMD), we sample representative chains from the Schulz-Zimm distribution and construct representative sets of large graphs encoding chain-scale topology directly, with atoms and bonds featurized using rich chemical descriptors. We further pretrain GNN encoders via masked graph modeling on 100,000 unlabeled PSMILES strings before fine-tuning on labeled data. On a dataset of 381 polymers (180 homopolymers and 201 copolymers), we show that graph construction and self-supervised pretraining are jointly necessary: without pretraining, the large graph method matches the repeat-unit baseline (28.40 K vs. 28.36 K RMSE); with pretraining, it achieves 24.76 K +/- 3.30 K, a 5.1% reduction in mean error over the pretrained repeat-unit baseline (26.08 K +/- 4.20 K, p < 0.001, 30 runs). An ablation removing chemical features degrades performance to 36.65 K, confirming both components are essential. Results are architecture-agnostic, holding for both GINE and GATv2 encoders.

Problem

• polymer prediction
• graph neural networks
• chain-scale topology
• molecular property prediction
• glass transition temperature

Innovation

• topology-aware graph encoding
• polymer property prediction
• self-supervised pretraining
• molecular mass distribution
• graph neural networks