Follow drug evidence across species.

Choose a target and filters

1. Choose a human target. The mapped proteins include exact ChEMBL single-protein targets for the human gene and indexed mouse orthologs.
2. Use the other controls to narrow by exact assayed protein, variant annotation, assay type, activity measurement, assay species, publication year or compound/publication text.
3. Read the snapshot summary before interpreting counts. The source query is bounded to up to 100 potency activities per target, sorted by pChEMBL, not by study date.
4. A matrix cell or activity bar filters activity records. A mechanism-action bar filters the separate mechanism table. A potency point highlights its activity row and opens its source context.
5. Download JSON to preserve the filtered records with coverage and source context, or TSV for spreadsheet work on activity rows.

What each graph counts

• Measurement types: sampled activity records grouped by exact assayed protein and measurement label. IC50, Ki, binding, inhibition and other labels are not interchangeable.
• Assay species: records grouped by the organism in which the assay was run. This is distinct from the target protein's species. The hatched unknown category means the source did not report an assay taxon.
• Variant annotation coverage: records with a variant annotation versus records without one. “Not specified” is missing annotation, not proof the assay used wild-type protein.
• Publication year: record counts by document year inside the current bounded sample. The sample was selected by pChEMBL value, so this is not a publication trend or a measure of research attention.
• Mechanism actions: ChEMBL mechanism annotations grouped by exact target and action. These are separate from assay activity records.
• Coverage funnel: per target, ChEMBL's count of potency-eligible activities, the rows retrieved from that window and the retrieved rows passing your filters. Rows from the mechanism-compound window are counted beside it.
• pChEMBL distribution (ECDF): for each target, the share of retrieved rows at or below each pChEMBL value. Further right is more potent. Human and mouse lines differ by dash and colour and are never pooled.
• Human and mouse side by side: rows, compounds and median pChEMBL for each species' own target. Nothing is transferred from one to the other.
• Maximum phase: distinct compounds by ChEMBL max phase. Phase belongs to the molecule, not the target. Choose a rung to filter the activity table.
• Compound bridge: compounds present in both the mechanism table and the activity sample. Choose one to filter both tables to it.
• Comparable potency: only exact, positive nM IC50, EC50, Ki or Kd records meeting the displayed source-quality rules appear. Points stay separated by target, assay species, variant, assay and endpoint; values are not averaged. Shape marks the endpoint, and the endpoint buttons hide or show each one. Drag across the plot, or focus it and use the arrow keys, to keep a pChEMBL range in the activity table; Shift with an arrow resizes the range and Escape clears it.

Coverage and missing values

Each target contributes at most 100 potency activities, highest pChEMBL first, plus activities for its mechanism compounds. Check the returned and total counts for truncated windows. No exact target match, no sampled activity, and a missing assay taxon or variant are different states; none establishes inactivity. Orthology does not transfer an activity between proteins. Read target species, assay species, variant, relation and units together. These records are not clinical evidence.

Collection limits, source checks, licensing and refresh steps are documented in the repository's Drug Evidence pilot guide.