metaRLK 2.0: An updated database of plant receptor-like kinases developed with structure- and deep learning-based functional annotation and classification.

Receptor-like kinases (RLKs) are the largest class of signal transduction proteins in plants and play crucial roles in adaptation to diverse ecological environments. The previously published metaRLK database contains 311 581 RLKs from 508 plant species; however, 13% lack clear domain or family assignments, and structural features are not systematically integrated. Here, we present metaRLK 2.0, a structure-enhanced RLK resource that integrates large-scale structure-based annotation, fold classification, network analysis, and semantic functional inference. Structural reannotation identified 677 distinct domain types, increasing the number of recognized types by 62.7%. CATH-based classification revealed that RLK domains are enriched in rigid β-fold architectures. A structural similarity network identified 70 newly defined RLK families, 50 of which are predicted to be associated with plant cell wall-related processes. Overall, 8% of previously unclassified RLKs were assigned to known or newly defined families. MetaRLK 2.0 provides an expanded, structure-informed platform for the exploration of RLK diversity, evolution, and structure-function relationships and serves as a comprehensive resource for future research in plant receptor biology.