Article: A topological deep learning framework for graph representation: application to metal-organic frameworks Journal: International Journal of Reasoning-based Intelligent Systems (IJRIS) 2026 Vol.18 No.13 pp.73 - 91 Abstract: The performance of metal-organic frameworks is critically determined by their topology, yet traditional methods struggle to automatically and accurately extract topological features. In this paper propose a novel topology-aware message-passing graph neural network that integrates persistent homology into the message passing process via a differentiable gating mechanism. This integration allows the network to dynamically modulate information flow along topologically critical paths, simultaneously capturing local chemical environments and global pore topology. We evaluate our model on two public datasets: the CoRE MOF 2019 dataset for topology classification and the quantum MOF database for CO<SUB align="right"><SMALL>2</SMALL></SUB> adsorption energy regression. The proposed approach achieves 87.5% classification accuracy (9.4% absolute improvement over the state-of-the-art graph attention network) and reduces the regression root mean square error to 0.25eV. By enabling more accurate prediction of MOF properties, this work can accelerate high-throughput screening and the discovery of high-performance porous materials. Inderscience Publishers - linking academia, business and industry through research

Title: A topological deep learning framework for graph representation: application to metal-organic frameworks

Authors: Kachen Zhang

Addresses: Department of Computer Science, Guizhou Police College, Guiyang, 550005, China

Abstract: The performance of metal-organic frameworks is critically determined by their topology, yet traditional methods struggle to automatically and accurately extract topological features. In this paper propose a novel topology-aware message-passing graph neural network that integrates persistent homology into the message passing process via a differentiable gating mechanism. This integration allows the network to dynamically modulate information flow along topologically critical paths, simultaneously capturing local chemical environments and global pore topology. We evaluate our model on two public datasets: the CoRE MOF 2019 dataset for topology classification and the quantum MOF database for CO₂ adsorption energy regression. The proposed approach achieves 87.5% classification accuracy (9.4% absolute improvement over the state-of-the-art graph attention network) and reduces the regression root mean square error to 0.25eV. By enabling more accurate prediction of MOF properties, this work can accelerate high-throughput screening and the discovery of high-performance porous materials.

Keywords: metal-organic framework; topological feature extraction; graph neural network; GNN; persistent homology; adsorption energy prediction.

DOI: 10.1504/IJRIS.2026.154058

International Journal of Reasoning-based Intelligent Systems, 2026 Vol.18 No.13, pp.73 - 91

Received: 10 Feb 2026
Accepted: 13 Mar 2026
Published online: 10 Jun 2026 *

Title: A topological deep learning framework for graph representation: application to metal-organic frameworks

Keep up-to-date