A research approach built on scientific rigour and epistemic honesty, where every association is weighted by the strength of its evidence, and no conclusion outruns its data.
My work sits at the intersection of epigenetics, genetic polymorphisms and metabolomics, with a focus on interpreting complex panels (genetic variants, organic-acid metabolic profiles, nutrigenetic and pharmacogenetic markers) to support individualised, nutrition-based decisions under medical supervision.
The guiding principle is simple: complex data are only useful when their uncertainty is made explicit. Rather than producing confident, prescriptive answers, the goal is to surface recurrent, statistically robust patterns, map them to biological pathways, and hand clinicians clearly caveated hypotheses they can test, never treatments to apply blindly.
A distinguishing feature of the method is that each link (between a variant, a metabolite, a pathway and a nutritional lever) is explicitly graded. This keeps well-supported findings visibly separate from speculation.
Backed by direct, replicated evidence in the relevant context.
Strong mechanistic rationale with supportive, if partial, evidence.
A reasonable inference that remains to be tested directly.
An exploratory idea, flagged as such and never acted on alone.
Genetic variants describe potential; metabolic panels describe what is actually happening. Reading them together is more informative than either alone. A core research aim is to integrate an individual's nutrigenetic and pharmacogenetic polymorphisms with their metabolomic phenotype, so that hypotheses account for personal variants.
Variants such as MTHFR/MTRR (folate and B12), COMT (catecholamines), DAO/HNMT (histamine) and detoxification genes (GST, SOD2, NQO1) set the individual's biochemical predispositions.
Organic-acid markers reveal the corresponding pathways in action, folate and B12 turnover, catecholamine and histamine metabolism, oxidative and mitochondrial status.
This genotype-informed calibration is evidence-graded and clinician-validated at every step. Reasoning over gene-variant × metabolite-pathway × evidence-strength is exactly where structured AI analysis adds value, and it is the direction of the MetaPlantAI project.
The near-term research stays deliberately focused on human metabolomics. But it sits inside a larger, measurable chain, a readout of health that connects the land to the person:
This framing rests on a defensible evolutionary principle, functional metabolic convergence: phylogenetically distant plants and microbes recurrently converge on similar bioactive modules, so different species can fulfil the same functional role. It grounds a bioactivity-based approach and opens a future line linking metabolism, plants and ecosystems under a One Health vision.
This research is developed within Associazione Virdimura (Palermo), a non-profit focused on integrated health, by an interdisciplinary team spanning biology, pharmacology, agronomy, plant pathology and clinical medicine. It also includes educational writing for a PANDAS/PANS patient association, a high-unmet-need cohort used illustratively, though the method is general.
The applied, AI-assisted expression of this approach is the MetaPlantAI project, submitted to Anthropic's AI for Science program (currently under evaluation).