Machine learning and supercomputer simulations predict interactions between gold nanoparticles and blood proteins

Researchers at the Center for Nanoscience at the University of Jyväskylä in Finland used machine learning and supercomputer simulations to investigate how tiny gold nanoparticles bind to blood proteins. This study discovered that favorable nanoparticle-protein interactions can be predicted from machine learning models trained from atomic-scale molecular dynamics simulations. A new methodology opens a way to simulate the effectiveness of gold nanoparticles as targeted drug delivery systems in precision nanomedicine.

Hybrid nanostructures between biomolecules and inorganic nanomaterials constitute a largely unexplored research field with potential for new applications in bioimaging, biosensing, and nanomedicine. The development of such applications relies heavily on understanding the dynamic properties of nanobiointerfaces.

Important processes such as electron charge transfer, chemical reactions, and remodeling of biomolecular surfaces can occur over a wide range of length and time scales, requiring atomistic simulations to be performed to model the properties of nanobiointerfaces. It is difficult to convert. In a suitable aqueous environment.

Machine learning helps study interactions at the atomic level

Researchers at the University of Jyväskylä recently demonstrated that atomic simulations of interactions between metal nanoparticles and blood proteins can be significantly sped up.

Based on extensive molecular dynamics simulation data of gold nanoparticles, protein-in-water systems, graph theory and neural networks were used to demonstrate the effects of nanoparticles on five common human blood proteins (serum albumin, apolipoprotein E). We have created a methodology that allows us to predict the most favorable binding sites. , immunoglobulin E, immunoglobulin G, fibrinogen). The machine learning results were successfully verified by long-time scale atomistic simulations.

“In recent months, we have performed calculations showing that it is possible to selectively target overexpressed proteins on the surface of cancer cells by functionalized gold nanoparticles carrying peptides and anticancer drugs. We also published our research,” said Hannu Hakkinen, professor of computational nanoscience.

“By using new machine learning techniques, we will investigate how drug-carrying nanoparticles interact with blood proteins and how those interactions alter the effectiveness of drug carriers. Now we can expand our research.”

Research will continue

The results will enable additional research to develop new computational methods to study interactions between metal nanoparticles and biomolecules.

“Machine learning is a very useful tool when considering the use of nanoparticles in diagnostic and therapeutic applications in the field of nanomedicine. This is one of our major goals,” says Hakkinen.

The research was published in two papers in the journals Advanced Materials and Bioconjugate Chemistry.

Computational resources were provided by the Finnish Grand Challenge projects BIOINT and NanoGaC on the LUMI and Mahti supercomputers hosted at the Finnish Supercomputing Center CSC.