Accurately predicting protein-protein interactions (PPIs) in dimeric complexes remains a fundamental challenge in computational biology. Although existing PPIs prediction models, such as AlphaFold-Multimer (AF-Multimer) and AlphaFold3 (AF3), have achieved impressive performance, they still suffer from unsatisfactory accuracy due to the limited availability of protein dimer structures, whose collection is both expensive and labor-intensive. Here, we introduce a simple yet effective pre-training method, termed split and merge proxy (SMP), that leverages abundant monomeric proteins to simulate various PPIs tasks for the first time. Specifically, SMP constructs pseudo-dimers by splitting monomer data into two subunits, referred to as pseudo-receptors and pseudo-ligands, and trains models to merge them back by predicting their pseudo interactions (e.g., contact or docking). This proxy task enables large-scale pre-training without additional cost. Models pre-trained with SMP and subsequently fine-tuned on real protein dimer datasets demonstrate consistently improved accuracy and generalization across multiple benchmarks, surpassing strong baselines. Notably, SMP delivers more accurate structure predictions than both AF-Multimer and AF3 on several CASP15 dimer targets. Our findings highlight SMP as a scalable strategy for harnessing monomeric data to advance protein complex modeling, providing insights into the linkage between monomers and multimers. Accurate prediction of protein-protein interactions is limited by the scarcity of high-quality complex structures. Here, authors introduce SMP, a strategy that leverages pseudo-dimers derived from monomers to improve accuracy and generalization across diverse protein interaction applications.

Chloromethane, a toxic gas primarily produced naturally, contributes to stratospheric ozone destruction. The anaerobic acetogen Acetobacterium dehalogenans can utilize chloromethane as a carbon and energy source, but the associated dehalogenase/methyltransferase has remained elusive. Through comparative transcriptomics we identify a gene cluster, cdmBCA, which encodes a corrinoid-dependent methyltransferase system distinct from the characterised Cmu system used for chloromethane degradation in aerobic methylotrophs. Biochemical characterization reveals that the Cdm system reacts with other haloalkanes, but not with methoxylated aromatics, unlike closely related O-demethylases. X-ray structural analysis of the protein CdmB shows a hydrophobic channelling system directing haloalkanes towards cobalamin-dependent activation. Homologous proteins are found in anaerobic prokaryotes, particularly within the phyla Bacillota and Asgardarchaeota, suggesting previously unidentified microbial transformation of chloromethane in the environment. Discovery of the Cdm dehalogenation/methyltransferase system sheds light on the microbial contribution to the global chloromethane cycle. Chloromethane, a toxic gas primarily produced naturally, contributes to ozone destruction. Here, the authors identify and characterise an enzyme system that dehalogenates chloromethane and appears to be specific to anaerobic microorganisms.

Small, nonpolar cyclic peptides can both bind challenging targets and cross cell membranes, making them attractive for addressing currently undruggable targets such as many protein–protein interactions (PPIs). However, developing such compounds de novo without prior information about lead structures such as natural ligands or fragments remains a notable challenge. Here we show that functional screening of structurally highly diverse cyclic peptide libraries synthesized at nanomole scale allows identification of sub-kDa inhibitors of a PPI. By screening 15,360 fully random cyclic peptides, we were able to identify an inhibitor of the E3 ligase adaptor Keap1 and its substrate Nrf2. Optimization by rapid design–build–test cycles produced a membrane-permeable compound active in live cells. This study demonstrates that large, diverse cyclic peptide libraries can enable the discovery of cell-permeable PPI inhibitors from the ground up, providing a way to harness the powerful modality of small cyclic peptides to address often difficult-to-target intracellular interactions. Membrane-permeable cyclic peptides offer access to difficult intracellular targets but discovery remains challenging. Here the authors show that cell-active cyclic peptides can be identified by screening sufficiently large and diverse libraries of small synthetic peptides.

Biomanufacturing offers a sustainable alternative to petrochemical manufacturing by harnessing engineered microorganisms to convert renewable feedstocks into diverse value-added products. Systems metabolic engineering has driven its advancement, enabling rational design of microbial strains with enhanced performance and robustness. Despite these achievements, large-scale commercialization remains limited, highlighting the persistent gap between laboratory success and industrial deployment. We examine the challenges and opportunities in advancing biomanufacturing towards industrial implementation through two cases, succinic acid and polyhydroxyalkanoates, that exemplify products nearing, yet not fully reaching, commercial maturity. Finally, we discuss how integrating artificial intelligence with systems metabolic engineering can accelerate sustainable biomanufacturing. Biomanufacturing holds promise as an alternative to the petrochemical industry, with benefits including increased sustainability as well as improved supply chain resilience, a factor of particular importance in times of geopolitical instability. Here the authors explore both the challenges and opportunities in progressing biomanufacturing from laboratory successes towards industrial-scale deployment.