The idea
Metabolites cannot be measured per cell type in most studies, but the genes that make, use, transport or sense them can. The Metabolic Integrator connects each measured metabolite to its reactions and genes in Human-GEM or Mouse-GEM, then asks which cell types express those genes and whether they change with condition.
Step by step
- Choose species and compartment. Mouse-GEM or Human-GEM; tissue, CSF or circulation. In plasma or CSF the tool reports uptake, use and sensing capacity — it never claims that a cell type is the source of a circulating metabolite.
- Load scRNA data. A cell-level matrix with metadata, a pseudobulk matrix, or a DE table.
- Load metabolomics. A sample × metabolite table with sample metadata. Missing values are never treated as zero; filtering, imputation and normalisation are explicit and recorded.
- Check identifiers. Validate inputs shows how each metabolite name, HMDB, KEGG or ChEBI ID was matched, which need review and which cofactors are set aside.
- Run and read the mechanism cards. Each card separates measured observations, statistical associations, biological hypotheses and suggested validation experiments.
Three analysis modes
- Matched animals — RNA and metabolites from the same animals: sample-level correlations are available.
- Unmatched groups — same conditions, different animals: condition effects are compared, paired correlations are disabled.
- Descriptive — no replicates: no p values are produced.
Open the Metabolic Integrator How it is validated
Questions
Does it prove that a cell type produces a metabolite?
No. It is a hypothesis generator: RNA-based scores are not metabolic flux. Each card suggests experiments to test the hypothesis.
Which reference networks are used?
Reduced versions of Human-GEM 2.1.1 and Mouse-GEM 1.8.0 (CC BY 4.0), plus a small cited supplement of receptor and transporter links.
Found something wrong or unclear? Write to contact@solveseverything.com. See also: how results are validated.