A Holobiont Knowledge Graph Prioritises Phytohormone-Mediated Mechanistic Routes from Arabidopsis Leaf Endophytes to Photosynthetic Parameters

Knowledge graph and reproducibility audit pipeline accompanying the manuscript "A Holobiont Knowledge Graph Prioritises Phytohormone-Mediated Mechanistic Routes from Arabidopsis Leaf Endophytes to Photosynthetic Parameters" (in preparation). This release contains: (1) Data/ — 15 Additional files: the complete normalised edge list (24,518 canonical edges, 13,358 nodes), per-phase subsets (Phases A–MU), four-hop mechanistic routes (1,119 routes: curated 545 / text-mined 574), five-hop routes (28,299), betweenness centrality rankings, entity-alias normalisation map, and null-model results (N1/N2/N3, 1,000 iterations). (2) code/audit/ — 10 Python scripts (t0_01 through ta_01) that regenerated the corrected results after three data defects were identified: entity-alias duplication (five synonymous IDs for ABA, two for JA), cross-layer gene identifier split, and AGI mis-annotation. The knowledge graph integrates 224-strain genome-scale metabolic models (i-At-LSPHERE, Schäfer et al. 2023) with the PlantConnectome 2025 host gene network to enumerate mechanistic routes from Arabidopsis leaf endophytes to photosynthetic parameters (Vcmax, Jmax, gs, gm, ETR) via phytohormone bridges.