To address these limitations, we introduce a novel framework: the Molecular Merged Hypergraph Neural Network (MMHNN). MMHNN innovatively incorporates a predefined set of molecular subgraphs, replacing each with a supernode to construct a compact hypergraph. This architectural change substantially reduces computational overhead while preserving essential molecular interactions.

Artificial intelligence (AI) is reshaping diverse fields in science, with molecular science being no exception. A recent study published in Research (Science Partner Journal) reports S²ALM, a powerful AI tool which uses a step-wise learning approach for a detailed analysis of antibodies. Using both 1D sequence and 3D structural data for antibody learning, the developed model outperforms prior models across key tasks, advancing antibody design for various diseases and redefining the future of therapeutic development.

To address the growing plastic crisis, researchers have developed a catalytic method that breaks down common plastic waste—polyethylene, polypropylene, and their mixtures—into valuable olefins used in fuel and chemical production. The process uses inexpensive base-metal catalysts and achieves up to 95% carbon recovery at just 320 °C—significantly milder than conventional thermal recycling. This energy-efficient strategy offers a promising alternative to traditional methods, which are energy intensive or rely on expensive noble-metal catalysts.