Canna·wiki
Method and sources
Canna·wiki is the cannabinoid encyclopaedia published by Phytogrammes. 452 molecules each have their own entry: structure, targets, affinity for the CB1 and CB2 receptors, French legal status. Access is free and requires no account.
What the index covers
Cannabinoids are not limited to THC and CBD. The index also covers the endocannabinoids the body makes, semi-synthetic cannabinoids, synthetic agonists and the pharmacological tools used to study them. A molecule appears here because it is documented, not because it is sold: most entries describe compounds that Phytogrammes does not sell, and several describe prohibited substances.
Sources and verification
Every entry cites its primary sources. The count behind this sentence is derived from the corpus, not copied over. Each entry gives the date of its last verification in its « Classification » chapter; the overall review of the corpus dates from .
Legal status changes faster than the rest: a molecule can be added to or removed from a ministerial order between two reviews. The date above says how far the review went, not that nothing has changed since. For a decision that commits you, the regulatory source cited in the entry is authoritative, never the entry itself.
Structures are checked, not drawn by hand
Every 2D diagram is drawn from the molecule's canonical SMILES. Before publication, the build derives the InChIKey from that SMILES and compares it with the one recorded in the corpus: if the two denote different molecules, the build fails and nothing is published. A discrepancy limited to stereochemistry is reported without blocking. This guard exists because the case occurred: one entry drew cannabicitran for months while its text described cannabitriol.
Correcting an entry
Every correctable section of an entry carries an invitation to propose a correction. Proposing requires an account, and the proposal is reviewed before publication: this is not open editing, it is a submitted correction. The index page itself is correctable, just as the entries are.
A useful correction cites its source. An affinity value, the date of an order or a CAS number is corrected against a reference; without one, the proposal cannot be settled.
How molecules are classified
Each molecule belongs to one family, defined by its origin: made by a plant, made by the body, transformed by the body (metabolites), obtained by modifying a hemp molecule (semi-synthetics), built in a laboratory to activate CB1 and CB2 receptors (synthetic agonists), or designed to block or tune the system without activating it (pharmacological tools). A sub-group locates it within the family. Among phytocannabinoids, « principal » means the four scaffolds that characterised enzymes make in the plant, CBG, THC, CBD and CBC, with their acids (Hanuš 2016; Maioli 2022); an origin code states whether a molecule is made by the plant or forms in it as it ages, like CBN. Medicine status and legal status are notes on the entry, not families.
Measured effects: the method
An entry's « Measured effects » section shows a level only when randomised, blinded, placebo-controlled human trials have measured the molecule with a validated instrument. Five constructs are used: psychoactive intensity, measured anxiety, cognitive impairment, sleepiness and reported hunger. Surveys, apps, case reports, and animal, in-vitro or receptor data never set a level.
The level follows from the difference from placebo at the peak effect, on a 100-point scale: under 10 points, not distinct from placebo; 10 to 19, weak; 20 to 39, moderate; 40 or more, marked. These cut-offs are an editorial convention, anchored on the gap of at least 15 points between placebo and positive control that the FDA requires to validate an abuse-potential study. An effect that changes direction with dose is shown as such; a significant effect whose size is not published is reported without a number. Confidence, from « high » to « very low », follows GRADE logic: two agreeing trials or one, then one step down for a small sample, no correction for multiple comparisons, industry funding, a contradicting study, or a data point found by only one of the two extractions.
The data were extracted twice, independently; a study on which the two extractions disagree is not used. Levels are computed by a script from these published extractions, so anyone can recompute them. Without a controlled trial, the entry shows « Not characterised »: that is the case for the great majority of molecules, and it is the right answer. These levels describe measurements, not benefits: none is an indication.
How entries are verified
Each entry is checked against its primary sources: identifiers verified on PubChem by InChIKey, dated and numerical claims compared with the cited publication, and every PMID resolved to make sure it is the paper named. The matching English Wikipedia article is paired by the molecule's structure, with a dated revision, and read as a second opinion: it helps spot what is missing or diverges, never as a source. Receptor binding values are no longer typed in by hand: each bar's length follows from the cited measured value (Ki, Kd, EC50 or IC50), on one scale shared by every entry.
What Canna·wiki is not
This is not medical advice, nor an invitation to consume. The entries describe a pharmacology and a legal framework; they recommend no use. For adults only. If in doubt, consult a healthcare professional.