Evidence Tier
computational
Last updated
Rung 7, the weakest rung on this site's ladder. A prediction produced by a model, such as molecular docking, structure prediction, QSAR or machine learning, with no wet-lab or clinical measurement of the claimed outcome. It qualifies only if the method and model are named and versioned, validation status is stated including not experimentally validated, and the prediction is displayed as a prediction. The full criteria are at /evidence/rubric.
Flagged for specialist review
This entry describes something contested, or asserts a detail we have not been able to confirm against a settled source. It is published rather than withheld, with the flag visible, because an entry you can weigh beats a gap you cannot see.
Why it matters here
The rung exists so predictions can be recorded honestly instead of quietly promoted, which is how a docking score ends up in marketing copy described as a measured affinity.
What it is not
A docking score is a hypothesis, and a confident-looking number from a model is not stronger than a rough measurement from a dish. A field reading not on file also sits below this rung entirely: it means no record meets any rung's criteria, which is not the same as a study having looked and found nothing.
Also called
- computational
- in silico
- model-predicted
- predicted
- docking study
In silico is the standard term, coined by analogy with in vivo and in vitro to mean in silicon, that is, in a computer.
Compounds carrying this term
None on file yet. This term is in the controlled vocabulary but no compound record currently uses it — which is a statement about our coverage, not about the term.