Evidence codes in the Atlas
Every study in the Atlas carries a code for the KIND of study it is, never a mark for how convincing it sounds — an M molecular paper can matter more to the field than an underpowered H human trial. The code describes what was studied, and nothing else. Until September 2026 these codes were the letters A–D; they were renamed because a lettered ladder reads as a school report no matter what the caption says. What changed and why.
The table below counts what the corpus is made of. For what those proportions mean — how much of the pathway reaches a human, how many links rest on a single paper, and how much of what is claimed carries a recorded boundary — see the evidence audit.
| Evidence | Studies | Share of coded studies |
|---|---|---|
| S Synthesis of human data | 3 | 1% |
| H Human study | 47 | 9% |
| A Animal model | 152 | 30% |
| M Molecular — cells, biochemistry, structure | 266 | 52% |
| R Review — secondary literature, not a new result | 43 | 8% |
What each code means
- S — synthesis of human data: a formal synthesis of multiple independent human studies (systematic review, meta-analysis).
- H — human study: a controlled or observational study in living humans.
- A — animal model: a whole-organism study in a non-human species (mouse, fly, worm, etc.).
- M — molecular: cell-culture, biochemical or structural work — usually where causal biology gets established first, well before anything reaches a human trial.
- R — review: secondary literature summarising other people's results rather than reporting a new one. Split out of the molecular bucket in September 2026, because a narrative review and a primary structural paper are not the same kind of document.
The code is independent of confidence: an M molecular finding can be extremely well-established (e.g. rapamycin directly inhibiting mTORC1), while an H human trial can still leave open questions about dose, duration or generalisability. Full criteria are on the About & Methodology page.
Thank you. Your answer was counted anonymously.