CANDI
About the project

A bridge between cannabis chemistry and human biology.

CANDI (Cannabis Target Identification) is a curated webserver that brings together cheminformatics, machine learning, and pathway biology to help researchers, formulators, and clinicians understand what a given cannabis compound — or an entire formulation — is most likely doing inside the human body.

Curated chemistry

Over 2,245 cannabis-relevant compounds annotated with SMILES, PubChem CIDs, and assay-derived bioactivity.

AI target prediction

Models trained on chemical-target interactions score the likely human protein targets for any compound.

Reactome pathways

Predicted targets are propagated to Reactome pathways for mechanistic interpretation.

How it works

  1. 1. Compound lookup
    Search by name, SMILES, or PubChem CID. Each compound is matched against the curated chemical library.
  2. 2. Target scoring
    An AI model scores affinity against a panel of human protein targets, ranked by predicted relevance.
  3. 3. Pathway enrichment
    Top targets are mapped onto Reactome pathways with normalized scoring across the formulation.
  4. 4. Similarity analysis
    Cosine similarity between compounds reveals shared mechanisms across the cannabinoid space.

Who is CANDI for?

Researchers
Generate target hypotheses for novel cannabinoids before benchwork.
Formulators
Predict the combined biological footprint of a multi-compound blend.
Clinicians
Anticipate pathway-level effects when reviewing cannabis-based regimens.

Contact

Dr. Nikolay Dokholyan, Professor

Departments of Neurology, Neuroscience, Pharmacology,
Biomedical Engineering, Microbiology, Immunology, & Cancer Biology
University of Virginia

dokhlab.org
Ready when you are
Start with a compound or your own formulation.
Search compounds Upload formulation