# Spectral Decomposition of Chemical Semantics for Activity Cliffs‐Aware Molecular Property Prediction

## TL;DR

PrismNet improves molecular property prediction by capturing chemical semantics and activity cliffs through a novel spectral decomposition approach.

## Contribution

Introduces PrismNet, a spectral graph network that decomposes molecules into multiple chemical perspectives for interpretable and accurate property prediction.

## Key findings

- PrismNet achieves state-of-the-art performance on 64 benchmark datasets, including 30 activity cliff datasets.
- The model autonomously identifies key substructures aligned with known structure-activity relationships.
- The dual-decomposition strategy captures both global topologies and subtle perturbations in molecular structures.

## Abstract

Accurately predicting physicochemical and biological properties of molecules is vital for modern drug discovery, yet existing deep learning models struggle to replicate the multi‐level reasoning of chemists. Relying on single molecular graphs, they fail to capture the interplay among global scaffolds, functional groups, and pharmacophoric patterns, and often miss subtle perturbations causing “activity cliffs”. PrismNet is proposed as a spectral graph network that mimics chemical intuition through a computational prism analogy. It applies a dual‐decomposition strategy: refracting molecules into three chemical perspectives—scaffolds, functional groups, and pharmacophores—and resolving each into spectral frequencies. A dynamic learning strategy further enhances its ability to handle heterogeneous data. PrismNet achieves state‐of‐the‐art performance across 64 benchmark datasets, including 30 activity cliff datasets. Importantly, its predictions are chemically interpretable, autonomously identifying key substructures aligned with known structure‐activity relationships. This framework unifies multi‐scale semantics and spectral decomposition, enabling reliable and trustworthy in silico screening for drug discovery.

PrismNet mimics chemical intuition by functioning as a computational prism, refracting molecular graphs into complementary semantic views and spectral frequencies. This dual‐decomposition strategy effectively captures both global topologies and subtle “activity cliff” perturbations. Achieving state‐of‐the‐art performance across 64 benchmarks, PrismNet provides an interpretable and robust solution for accelerating data‐driven drug discovery.