Machine-learned potentials (MLPs) have become widely adopted alternatives to traditional electronic structure and molecular mechanics methods; however, most MLPs remain prone to instability when deployed in molecular dynamics simulations, particularly at elevated temperatures. Here, the authors present physics-informed Gaussian process (GP)-based atomic energy models that achieve enhanced stability in NVT simulations at temperatures as high as 1000 K, demonstrating their robustness in simulations of flexible organic molecules (peptide-capped glycine and serine, malondialdehyde, and aspirin) with a cumulative simulation time of 0.5 microseconds completed within two CPU days.

Disulfide tethering screening is applied to address the challenge of targeting oncogenic KRAS, historically deemed undruggable due to its high nucleotide affinity and perceived lack of binding sites. Here, the authors developed a disulfide library and a set of 83 cysteine KRAS mutants to perform tethering screens towards the oncogenic KRAS, while describing the discovery of fragment-binding pockets in KRAS G12D, identified by this screening approach.

While Δ⁹-tetrahydrocannabinol (THC) offers significant pain relief, its psychoactive side effects limit its clinical utility. Here, the authors explore cannabidiphorol (CBDP) as a dual-site negative allosteric modulator of cannabinoid receptor 1, revealing its potential to modulate cognitive effects while preserving THC-induced antinociception.

Tetrahydroisoquinoline scaffolds are crucial in natural products and pharmaceuticals due to their diverse biological activities, yet synthetic pathways for their polycyclic forms are limited. Here, the authors present a metal-free, Vilsmeier-reagent promoted decarbonylative cyclization method, yielding derivatives with notable antitumor activity.

High-fold degenerate topological semimetals with chiral structures are promising candidates for catalyzing the hydrogen evolution reaction (HER). Here, the authors predict 16 high-activity topological catalysts with Gibbs free energies smaller than that of Pt, showing that long topological surface states can improve HER catalytic performance, particularly PtGa and PtPbTe.

Machine learning models for the prediction of molecular properties are often trained on relatively scarce data, resulting in memorization — rather than learning — and poor generalizability. Here, the authors develop a pKa predictor based on a teaching-based approach, demonstrating that the same fundamental principles that are taught to students — such as electronegativity, conjugation and aromaticity — can be used to train their machine learning model, which is shown to outperform state-of-the-art predictors.

Mechanoradicals are known to be sources of damage during polymer processing, and it remains challenging to stabilize, repair or utilize them. Recently, mechanoradicals formed during chain scission were intercepted directly by bis(butyl trithiocarbonate) in the solid state, redirecting the degradation pathways to repolymerization (repair) and complete depolymerization (recovery).

We are pleased to present our Editors’ Highlights Collection, featuring some of our favourite content published in Communications Chemistry in 2025. Here, we highlight why each paper was selected by our Editors and Editorial Board Members.

Mental health should be a priority for all and especially for underrepresented groups, including the women in chemistry. Here, the authors discuss and propose supportive mechanisms within academia for the mental health of undergraduate women in chemistry.

Dr Laura Cattaneo is a group leader at the Max Planck Institute for Nuclear Physics in Heidelberg, Germany. She has an interdisciplinary research background in chemical physics, connecting materials science, liquid crystals, and ultrafast processes in matter.

Electrocatalytic nitrate reduction is widely studied, but under a range of conditions, with implications on catalyst performance. Here, the authors highlight potential pitfalls in comparing across (and within) existing literature and suggest approaches to assess electrocatalysts under common conditions.

Communications Chemistry is pleased to present a Collection of articles on triplet-harvesting materials. Here, the Guest Editors introduce the topic and look towards the future of the field.