A Compound-Centric Framework for Mapping Plant Chemical Space through Structural Scaffolds and Bioactivity Evidence

Open repositories have greatly expanded access to plant chemical occurrence data at scale, yet their direct use in compound prioritization remains limited by taxonomic inconsistency, structural redundancy, and uneven bioactivity annotation. Here, we present a compound-centric framework that transforms literature-curated occurrence records into interpretable, evidence-aware outputs by integrating LOTUS species-structure associations, World Flora Online taxonomic reconciliation, InChIKey-based dereplication, Murcko scaffold analysis, and conservative ChEMBL-linked bioactivity evidence. Applied to Phyllanthaceae as a case study, the workflow organizes reported chemical space across genera, resolves structure-level redundancy, and links scaffold-level novelty to family-level rarity and indexed activity records. This strategy distinguishes two complementary evidence-aware categories: lineage-restricted compounds with ChEMBL-linked activity records (STARS) and lineage-restricted compounds lacking qualifying ChEMBL activity records (HIDDEN GEMS). More broadly, the taxon-flexible design supports analyses across multiple taxonomic scales in plant lineages represented in LOTUS. It provides an auditable framework for converting heterogeneous occurrence data into transparent outputs that support evidence-aware prioritization in natural products research.