A recent study, “Meta-Analysis and Machine Learning Prediction of Protein Corona Composition across Nanoparticle Systems in Biological Media,” by Wei-Chun Chou and colleagues at the University of California, Riverside (USA), integrates two decades of nanoparticle-protein corona research to better understand how nanoparticle properties dictate protein adsorption patterns.

Key Findings:
• The team built the Protein Corona Database (PC-DB), integrating data from 83 studies, 817 nanoparticle formulations, and 2,497 proteins across four species.
• Meta-analysis revealed that silica, polystyrene, and lipid-based nanoparticles <100 nm with near-neutral ζ-potentials preferentially adsorb APOE and APOB-100, proteins associated with receptor-mediated uptake.
• Metal and metal-oxide nanoparticles with highly negative surface charge preferentially enriched complement component C3, indicating a greater likelihood of immune recognition.
• Interpretable ML models (LightGBM, XGBoost) achieved ROC-AUC > 0.85, identifying size, ζ-potential, and incubation time as the strongest predictors of protein adsorption.

Why It Matters:
This study introduces a unified, open-access resource that enables scientists to predict protein corona profiles from nanoparticle features. Such predictive capability accelerates rational nanocarrier design, improving control over biodistribution, immune interactions, and delivery efficiency.

Explore the publication: https://doi.org/10.1021/acsnano.5c08608

Access the Protein Corona Database (PC-DB): https://wb.ai2tox.com