Summary
A bioRxiv preprint presents a physics-based framework that combines protein, nucleic-acid and metabolite models to simulate biomolecular condensates. Its authors report predictions for more than 60 disordered proteins and simulations of RNA- and protein–nucleic-acid condensates.
A bioRxiv preprint describes a physics-based framework for simulating how proteins, nucleic acids and cellular metabolites form biomolecular condensates. The framework combines the HyRes protein model and iConNA nucleic-acid model, and adds a machine-learning-enabled metabolite library called iConMetabolome. The authors report predictions of saturation concentrations for more than 60 disordered proteins, as well as simulations of RNA and mixed protein–nucleic-acid condensates.
Biomolecular condensates are concentrated assemblies that can form when molecules with multiple interaction sites separate from their surroundings into a distinct phase. They help organise cellular components. Simulating their formation requires representing the interactions that draw molecules together and those that keep the resulting assemblies dynamic.
A shared framework for several molecular interactions
The framework represents hydrogen bonding, electrostatic forces, cation–π interactions and hydrophobic interactions. For nucleic acids, it also includes base stacking, base pairing and effects mediated by ions. This lets the model treat proteins and nucleic acids within a shared simulation approach rather than modelling each type of condensate with a separate framework.
The authors report that the model predicts saturation concentrations for more than 60 disordered proteins within tenfold. Saturation concentration is the concentration threshold associated with the onset of phase separation. They also report phase diagrams for homotypic RNA condensates—formed from RNA—and heterotypic condensates containing proteins and nucleic acids. A phase diagram describes the conditions under which different phases form.
The metabolite component extends the framework to small molecules found in cells. The authors say iConMetabolome captures metabolite-driven phase separation and reproduces experimental trends in how metabolites partition across condensates. Partitioning describes how a molecule distributes between a condensate and its surrounding solution.
What the framework could help investigate
Combining these molecular components could help researchers examine how interaction types and metabolites influence condensate formation. The authors present the platform as a way to investigate the mechanisms and regulation of condensation in biology and disease. Its stated scope is modelling condensates and their molecular interactions, not simulating an entire cell.
The work is a bioRxiv preprint, and the reported findings concern computational predictions and comparisons with experimental measurements or trends. The abstract provides headline performance results but not detailed validation methods, so the reported accuracy figures should be understood at the level described there.