%0 Journal Article
%T High-Content Phenotypic Imaging as a Computational Discovery Space for Linking Natural Product Chemistry to Cellular Morphology and Mechanism of Action
%A João Silva
%A Pedro Costa
%A Ana Beatriz
%J International Journal of Pharmaceutical And Phytopharmacological Research
%@ 2250-1029
%D 2025
%V 15
%N 1
%R 10.51847/iBKm6KDcw1
%P 67-76
%X Natural products occupy chemically diverse and often sparsely annotated regions of bioactive chemical space, where a single target assay can capture only a narrow fraction of their cellular effects. High-content phenotypic imaging offers a complementary strategy by representing drug-induced cellular responses as multidimensional morphological signatures that can be searched, compared and modeled computationally. This conceptual framework examines how natural-product chemistry can be connected to image-derived cellular morphology without treating phenotypic similarity as direct evidence of molecular mechanism. It distinguishes the measured image, the computational representation derived from it, the morphology–chemistry relationship inferred by models, and the orthogonal evidence required for mechanism-of-action attribution. The framework proposes a context-indexed morphological discovery space in which compound identity, dose, exposure time, cell state, imaging protocol and representation method jointly determine the interpretable phenotype. It further positions reference compounds, chemical structures, transcriptomic or proteomic measurements and target-focused experiments as complementary evidence layers rather than interchangeable validators. Morphological similarity can therefore prioritize mechanistic hypotheses, identify activity in chemically underannotated space and guide which natural products merit deeper testing, but it can also arise from cytotoxicity, stress, convergent downstream biology or technical confounding. The proposed architecture is intended as a discovery logic rather than a validated inference rule. Its utility depends on reproducible profiling, domain-aware modeling, explicit uncertainty and orthogonal confirmation before target-level or translational conclusions are drawn.
%U https://eijppr.com/article/high-content-phenotypic-imaging-as-a-computational-discovery-space-for-linking-natural-product-chemi-ffchywz2xti4omb