English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Protein (receptor)--ligand interaction prediction is a critical component incomputer-aided drug design, significantly influencing molecular docking andvirtual screening processes. Despite the development of numerous scoringfunctions in recent years, particularly those employing machine learning,accurately and efficiently predicting binding affinities for protein--ligandcomplexes remains a formidable challenge. Most contemporary methods aretailored for specific tasks, such as binding affinity prediction, binding poseprediction, or virtual screening, often failing to encompass all aspects. Inthis study, we put forward DeepRLI, a novel protein--ligand interactionprediction architecture. It encodes each protein--ligand complex into a fullyconnected graph, retaining the integrity of the topological and spatialstructure, and leverages the improved graph transformer layers with cosineenvelope as the central module of the neural network, thus exhibiting superiorscoring power. In order to equip the model to generalize to conformationsbeyond the confines of crystal structures and to adapt to molecular docking andvirtual screening tasks, we propose a multi-objective strategy, that is, themodel outputs three scores for scoring and ranking, docking, and screening, andthe training process optimizes these three objectives simultaneously. For thelatter two objectives, we augment the dataset through a docking procedure,incorporate suitable physics-informed blocks and employ an effectivecontrastive learning approach. Eventually, our model manifests a balancedperformance across scoring, ranking, docking, and screening, therebydemonstrating its ability to handle a range of tasks. Overall, this researchcontributes a multi-objective framework for universal protein--ligandinteraction prediction, augmenting the landscape of structure-based drugdesign.

DeepRLI: A Multi-objective Framework for Universal Protein--Ligand Interaction Prediction