Weiqi Zhang
Weiqi Zhang has a total of 28 co-authors. Here are the top co-authors by number of shared publications:
- Maya Thanou
School of Cancer and Pharmaceutical Sciences, King’s College London
4 shared publications - Stavros Vlatakis
Institute of Pharmaceutical Science, King’s College London, London SE1 9NH, UK
3 shared publications - Hilde Metzger
James Watt School of Engineering, University of Glasgow, Glasgow, UK
2 shared publications Aaran Vijayakumaran
School of Cancer & Pharmaceutical Sciences, King's College London, UK
1 shared publication- Alexandru Corneliu Moldovan
James Watt School of Engineering, University of Glasgow, Glasgow G12 8QQ, UK
1 shared publication - Amelia Claxton
Innovation Hub, Comprehensive Cancer Centre, King’s College London, Great Maze Pond, London SE1 9RT, UK
1 shared publication
Latest Publications
- Open Access
- 9 November 2026
- Published by American Chemical Society (ACS) in ChemRxiv
Chemistry-informed non-covalent graphs and pharmacophore supervision for molecular property prediction
Graph neural networks describe a molecule by its covalent bonds, however, many molecular properties are shaped by contacts between atoms that are not bonded to each other. Existing models either ignore these non-covalent interactions or approximate them with a distance cutoff that cannot distinguish an interaction from mere proximity. Here we introduce a framework that injects chemical knowledge at two levels. First, intramolecular hydrogen bonds and van der Waals contacts become graph edges only when they satisfy interaction-specific topological and geometric criteria; these edges form a separate graph whose layer refines the atom states of a covalent encoder, and the two readouts are fused through a task-aware gate. Second, an auxiliary head supervises every atom with its pharmacophoric roles with expanded σ-hole interaction dictionary. We show that the rule-based graph is about twenty times sparser than a 5-Å cutoff graph, that the gated model reduces to a covalent model as its gate closes and adds a generalisation term bounded by a multiple of its maximal gate value, and that its predictions are Lipschitz in the atomic coordinates as long as the edge set is unchanged. On seven MoleculeNet classification benchmarks under scaffold splitting and without pretraining, ChemPrior-GAT achieves the best ROC-AUC on three benchmarks and the lowest average rank, and the same two modules also improve AttentiveFP, demonstrating that the proposed strategy is transferable to other GNN architecture. Pharmacophore-level attributions are consistent with established chemical principles. In prospective experimental validation involving 16 compounds, the model achieved a mean absolute error (MAE) of 0.50 log units between predicted and experimentally measured logD7.4 values.
- 9 November 2026
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessChemistry-informed non-covalent graphs and pharmacophore supervision for molecular property prediction
Graph neural networks describe a molecule by its covalent bonds, however, many molecular properties are shaped by contacts between atoms that are not bonded to each other. Existing models either ignore these non-covalent interactions or approximate them with a distance cutoff that cannot distinguish an interaction from mere proximity. Here we introduce a framework that injects chemical knowledge at two levels. First, intramolecular hydrogen bonds and van der Waals contacts become graph edges only when they satisfy interaction-specific topological and geometric criteria; these edges form a separate graph whose layer refines the atom states of a covalent encoder, and the two readouts are fused through a task-aware gate. Second, an auxiliary head supervises every atom with its pharmacophoric roles with expanded σ-hole interaction dictionary. We show that the rule-based graph is about twenty times sparser than a 5-Å cutoff graph, that the gated model reduces to a covalent model as its gate closes and adds a generalisation term bounded by a multiple of its maximal gate value, and that its predictions are Lipschitz in the atomic coordinates as long as the edge set is unchanged. On seven MoleculeNet classification benchmarks under scaffold splitting and without pretraining, ChemPrior-GAT achieves the best ROC-AUC on three benchmarks and the lowest average rank, and the same two modules also improve AttentiveFP, demonstrating that the proposed strategy is transferable to other GNN architecture. Pharmacophore-level attributions are consistent with established chemical principles. In prospective experimental validation involving 16 compounds, the model achieved a mean absolute error (MAE) of 0.50 log units between predicted and experimentally measured logD7.4 values.
- Open Access
- 28 December 2023
- Published by MDPI AG in Pharmaceutics
- Vol. 16 (1) , 51
Effect of Phase-Change Nanodroplets and Ultrasound on Blood–Brain Barrier Permeability In Vitro
Phase-change nanodroplets (PCND;NDs) are emulsions with a perfluorocarbon (PFC) core that undergo acoustic vaporisation as a response to ultrasound (US). Nanodroplets change to microbubbles and cavitate while under the effect of US. This cavitation can apply forces on cell connections in biological barrier membranes, such as the blood–brain barrier (BBB), and trigger a transient and reversible increased permeability to molecules and matter. This study aims to present the preparation of lipid-based NDs and investigate their effects on the brain endothelial cell barrier in vitro. The NDs were prepared using the thin-film hydration method, followed by the PFC addition. They were characterised for size, cavitation (using a high-speed camera), and PFC encapsulation (using FTIR). The bEnd.3 (mouse brain endothelial) cells were seeded onto transwell inserts. Fluorescein with NDs and/or microbubbles were applied on the bEND3 cells and the effect of US on fluorescein permeability was measured. The Live/Dead assay was used to assess the BBB integrity after the treatments. Size and PFC content analysis indicated that the NDs were stable while stored. High-speed camera imaging confirmed that the NDs cavitate after US exposure of 0.12 MPa. The BBB cell model experiments revealed a 4-fold increase in cell membrane permeation after the combined application of US and NDs. The Live/Dead assay results indicated damage to the BBB membrane integrity, but this damage was less when compared to the one caused by microbubbles. This in vitro study shows that nanodroplets have the potential to cause BBB opening in a similar manner to microbubbles. Both cavitation agents caused damage on the endothelial cells. It appears that NDs cause less cell damage compared to microbubbles.
- 28 December 2023
- Published by MDPI AG in Pharmaceutics
- Vol. 16 (1) , 51
Open AccessEffect of Phase-Change Nanodroplets and Ultrasound on Blood–Brain Barrier Permeability In Vitro
Phase-change nanodroplets (PCND;NDs) are emulsions with a perfluorocarbon (PFC) core that undergo acoustic vaporisation as a response to ultrasound (US). Nanodroplets change to microbubbles and cavitate while under the effect of US. This cavitation can apply forces on cell connections in biological barrier membranes, such as the blood–brain barrier (BBB), and trigger a transient and reversible increased permeability to molecules and matter. This study aims to present the preparation of lipid-based NDs and investigate their effects on the brain endothelial cell barrier in vitro. The NDs were prepared using the thin-film hydration method, followed by the PFC addition. They were characterised for size, cavitation (using a high-speed camera), and PFC encapsulation (using FTIR). The bEnd.3 (mouse brain endothelial) cells were seeded onto transwell inserts. Fluorescein with NDs and/or microbubbles were applied on the bEND3 cells and the effect of US on fluorescein permeability was measured. The Live/Dead assay was used to assess the BBB integrity after the treatments. Size and PFC content analysis indicated that the NDs were stable while stored. High-speed camera imaging confirmed that the NDs cavitate after US exposure of 0.12 MPa. The BBB cell model experiments revealed a 4-fold increase in cell membrane permeation after the combined application of US and NDs. The Live/Dead assay results indicated damage to the BBB membrane integrity, but this damage was less when compared to the one caused by microbubbles. This in vitro study shows that nanodroplets have the potential to cause BBB opening in a similar manner to microbubbles. Both cavitation agents caused damage on the endothelial cells. It appears that NDs cause less cell damage compared to microbubbles.
- Open Access
- 17 May 2023
- Published by Elsevier in Ultrasonics Sonochemistry
- Vol. 97, 106445
Characterising the chemical and physical properties of phase-change nanodroplets
- 17 May 2023
- Published by Elsevier in Ultrasonics Sonochemistry
- Vol. 97, 106445
Open AccessCharacterising the chemical and physical properties of phase-change nanodroplets
- 15 February 2022
- Published by Royal Society of Chemistry (RSC) in Nanoscale
- Vol. 14 (8) , 2943-2965
Phase-shift nanodroplets as an emerging sonoresponsive nanomaterial for imaging and drug delivery applications
The scheme of loading drugs/decorations in nanodroplets and US/laser irradiated drug release (figure created with BioRender.com).
- 15 February 2022
- Published by Royal Society of Chemistry (RSC) in Nanoscale
- Vol. 14 (8) , 2943-2965
Phase-shift nanodroplets as an emerging sonoresponsive nanomaterial for imaging and drug delivery applications
The scheme of loading drugs/decorations in nanodroplets and US/laser irradiated drug release (figure created with BioRender.com).
- 23 November 2019
- Published by Springer Nature
- p. 205-239
Design of Nanoparticles for Focused Ultrasound Drug Delivery
- 23 November 2019
- Published by Springer Nature
- p. 205-239
Design of Nanoparticles for Focused Ultrasound Drug Delivery