Fig. 1. Overview of RE3DB data integration and interactive modules. RE3DB integrates curated rice E3 ubiquitin-ligase-system gene annotations, protein-protein interaction (PPI) resources, RNA sequencing (RNA-seq) expression profiles, protein-to-transcript ratio (PTR) data, and family-level phylogenies. The platform contains three interactive modules: classification and annotation, phylogeny-guided redundancy prioritization, and E3 interaction prediction. Main outputs include family-level summaries, downloadable gene annotation tables, phylogenetic heatmaps, and E3-centered interaction networks. RING, Really Interesting New Gene; HECT, Homologous to E6AP C-terminus; CRL, Cullin-RING ligase; RGAP, Rice Genome Annotation Project; SRA, Sequence Read Archive; PCC, Pearson correlation coefficient.

Fig. 2. Classification and annotation of rice E3-system genes in RE3DB. A, Word-cloud view of curated rice E3-system families. Font size is proportional to the number of genes in each family. B, Family-specific annotation table showing Rice Genome Annotation Project (RGAP, MSU7) locus identifiers, E3 family and subfamily, functional role category, Pfam-based description, Pfam entry, and InterPro entry. Hyperlinked locus identifiers connect to RGAP gene pages, and Pfam or InterPro entries connect to the corresponding domain pages. OsBBI1 (BLAST AND BTH-INDUCED 1; LOC_Os06g03580) is highlighted as an example. The download button exports the displayed table as a CSV file. APC, Anaphase-promoting complex; BTB, Broad-complex, tramtrack and bric-à-brac; CDC20, Cell division cycle 20; DDB1, DNA damage-binding protein 1; DWD, DDB1-binding WD40; HECT, Homologous to E6AP C-terminus; RBX1, RING box protein 1; RING, Really Interesting New Gene; SKP1, S-phase kinase-associated protein 1; TAD1, Tiller Angle Control 1.

Fig. 3. Phylogenetic heatmaps for redundancy-aware prioritization of rice E3-system genes. A, RIP1-RIP6 RING-type E3 clade, a previously validated redundant gene set. B, GW2 (LOC_Os02g14720), a RING-type E3 gene known to act as a major grain-size regulator. C, LOC_Os10g04370, a mature-anther-preferential F-box candidate prioritized as a single-locus exploratory example. D, LOC_Os03g49250, a mature-anther-preferential F-box candidate whose local clade includes LOC_Os04g31580 as a possible companion paralog for future testing. RE3DB overlays transcript abundance across 25 rice tissues onto family-level protein phylogenies. For single-gene queries, Pearson correlation coefficients (PCCs) and raw two-tailed Pearson correlation P-values are shown for neighboring genes relative to the query locus.

Fig. 4. Integrated evidence for E3 interaction prediction in RE3DB. A, Querying OsFBK16 (LOC_Os06g39370) returns predicted interacting proteins, including OsPAL1 (LOC_Os02g41630) and OsPAL2 (LOC_Os02g41650), which were previously supported as OsFBK16-associated proteins. B, Querying TAD1/TE (LOC_Os03g03150), a Cdh1/CCS52-type anaphase-promoting complex (APC) activator, returns OsCDC27 (LOC_ Os06g41750), an APC core subunit, as a multi-source-supported E3-complex interaction. RE3DB predicts E3-centered interactions by combining high-confidence protein-protein interaction (PPI) resources, transcript co-expression across 25 tissues, raw Pearson correlation coefficient (PCC)-associated P-values, and low protein-to-transcript ratio (PTR) information across 14 tissues. The output table reports supporting PPI sources, source-specific scores, PCC, raw P-value, and tissues with low PTR, rather than converting heterogeneous source scores into a single composite score.

Fig. 5. Enrichment of tissue-preferential E3-system families. For each E3-system family, the top enriched anatomical terms ranked by odds ratio and passing false-discovery-rate (FDR)-adjusted P ≤ 0.05 are shown. Dot size indicates the percentage of tissue-preferential family members assigned to each term, and dot color indicates the odds ratio. Because E3-system families differ greatly in size, enrichment should be interpreted using odds ratios and FDR-adjusted P-values rather than raw gene counts alone. Complete tissue-preferential gene lists are provided in Tables S6 and S7, and family-specific expression profiles are shown in Figs. S3-S12. RING, Really Interesting New Gene.

Fig. 6. Prioritization of mature-anther-preferential F-box candidates associated with rice pollen germination. A, UpSet plot of mature-anther-preferential F-box genes that were stringently downregulated, defined as log2(fold change) ≤ -2, in nine male-sterile mutant backgrounds affecting pollen germination or pollen-tube growth. Matrix dots indicate the mutant backgrounds included in each intersection group. Vertical cylinders show the number of candidate genes in each intersection, and horizontal bars show the total number of candidates in each mutant set. Distinct cylinder colors identify intersection groups; they do not indicate fold-change magnitude or statistical significance. Matching hex color codes are provided in Table S10. B, Hypothesis-generating network integrating stringently downregulated F-box candidates with literature-curated pollen-germination regulators. Solid edges indicate experimentally supported regulatory relationships or protein-protein interactions, whereas dotted edges indicate inferred associations based on shared downregulation patterns in RNA-seq data.