A multimodal fusion framework integrating sequence, atomic, and fragment representations captures drug–target interactions across multiple scales. The model delivers strong predictive performance and enables efficient virtual screening. Applied to hematopoietic progenitor kinase 1 (HPK1), it identifies structurally diverse inhibitors with nanomolar ...
Shuo Liu +7 more
wiley +1 more source
(i) Aging impairs peripheral nerve regeneration via its dual ferroptosis‐mediated pathologies in senescent schwann cells. (ii) The composite nerve conduit is a symbiosis niche integrating microenvironment‐responsive nanoparticles (Ga‐PTAs) and maxillofacial‐derived mesenchymal stem cells (Mmscs).
Ning Zhan +12 more
wiley +1 more source
Robots and Minimal, Physics‐Informed Features: A Hybrid Framework for Enzyme Catalysis
Robotic experimentation and physics‐informed machine learning combine to predict enzyme substrate scope. With a handful of interpretable features derived from docking and quantum mechanics calculations, our model rivals descriptor‐heavy AI approaches and extrapolates to unseen substrates and enzyme classes.
Natalia Onishchenko +8 more
wiley +1 more source
DyProL: Dynamic Ensemble Representation Learning for Protein–Nucleic Acid Binding Site Prediction
Protein function is represented as a dynamic conformational ensemble rather than a single static structure. A multi‐conformation geometric attention framework aligns, clusters, and learns representative states to capture residue‐ and ensemble‐level signals. Integrating structural dynamics improves interpretable protein‐NA binding prediction and reveals
Pengpai Li +3 more
wiley +1 more source
VG-RETFound: a hybrid retinal AI framework for early vision impairment screening through vessel-graph learning and cross-modal fusion. [PDF]
Dutta AK, Aljarallah NA, Wahab Sait AR.
europepmc +1 more source
TECTB Variants Reveal Tectorial Membrane Vulnerability in Dominant Non‐Syndromic Hearing Loss
TECTB is a non‐collagenous protein of the tectorial membrane – an extracellular matrix of the cochlea. This study identifies dominant missense variants in TECTB linked to human hereditary deafness in two unrelated families. Genetically engineered mice homozygous for one of the variants are profoundly deaf, whereas heterozygous mice have normal hearing ...
Evan B. Hale +23 more
wiley +1 more source
Perceptual no-reference image quality assessment with meta-learning by graph representation learning and multi-scale feature fusion. [PDF]
Jia Y, Wei L.
europepmc +1 more source
Combined structure‐function assessment in aortic rings from Marfan mice using MechaMorph Opto‐Biomechatronics technology relates increased elasticity (stiffness) and dynamic viscosity and less ordered extracellular matrix 3D‐structure (collagen) as potential causes for impaired Windkessel function and compromised haemodynamics in Marfan's syndrome ...
Dominik Schneidereit +10 more
wiley +1 more source
Automated Tissue Classification and Candidate Biomarker Feature Extraction in Mass Spectrometry Imaging Based on Interpretable Deep Learning Algorithm. [PDF]
Xu G, Gan S, Guo B, Yang L.
europepmc +1 more source
Influence of Top Electrode Metal on Resistive Switching in Multilayer MOCVD MoS2 Memristors
MoS2 memristors typically use evaporated noble metal electrodes. This study compares Pd, Ni, and Al electrodes and sputtered versus evaporated deposition. Pd is passive, while Ni damages MoS2, yielding poor switching. Sputtered Al enables reproducible ECM‐type switching with 95% yield, whereas evaporated Al forms an interfacial oxide that suppresses ...
Dennis Braun +15 more
wiley +1 more source

