Medicinal-plant pathway discovery increasingly combines genome assemblies, gene-expression profiles, metabolite measurements, spatial information, and functional assays, yet the presence of multiple data layers does not itself resolve which gene produces, modifies, transports, or regulates a given metabolite. This article develops an original, evidence-bounded multi-omics architecture for gene–metabolite assignment in specialized metabolism. The analysis separates candidate nomination from biological and chemical constraint, conflict diagnosis, and functional resolution. It argues that genomics is most informative for candidate space, locus organization, duplication, and evolutionary context; transcriptomics for conditional coordination and cellular context; metabolomics for chemical occurrence and transformation plausibility; and spatial or paired-cell measurements for localization and cross-modality alignment. Agreement among layers strengthens an assignment only when the observations are sufficiently independent and mechanistically compatible. Conversely, disagreement should not automatically invalidate a candidate because transport, compartmentation, paralogy, pathway branching, analytical incompleteness, and reference-genome defects can generate legitimate discordance. The proposed architecture therefore treats confidence as claim-specific and conditional on evidence quality, orthogonality, compatibility, and the degree of functional testing, rather than on the number of omics layers supporting a candidate. Its intended output is not a universal score but an explicit, revisable statement of what is supported, what remains alternative, and what validation would be most discriminating. The framework is limited by uneven medicinal-plant resources, incomplete metabolite identification, context-dependent measurements, and the absence of prospective benchmarking across diverse pathways.
