StructureMoa is a chemical-structure exploration tool. Structural similarity, clustering, and SAR relationships are intended for research exploration and do not establish shared potency, selectivity, safety, efficacy, or clinical performance.

구조 탐색 연산 · 신경

Catalog graph and stored Morgan fingerprints: sparse SAR, cross-scaffold Tanimoto, leaf nodes, weakly linked compounds, and approved-tag gaps. No new candidates are invented.

제안 6화합물 9413 패밀리

Leaf node3

Few derived children in the catalog graph.

Approved-tag gap3

Few mid-similarity bridges around an approved or clinical-tagged drug.

Exploration score method

StructureMoa ranks are a rule-based Exploration Score. An LLM does not predict activity or efficacy. Role, SMILES, and relation signals inside each family are summed and cut into High / Medium / Low exploration priority.

Core tab

  • scaffold / payload / linker role weights
  • family center-compound bonus
  • SMILES source (db · known · pubchem · inherited)
  • relation-edge count and derived-child count

Derivative tab

  • drug / derivative / adc candidates
  • similarity to family center via Morgan fingerprint, radius 2, 2048 bits, Tanimoto
  • approved / clinical catalog-tag bonuses
  • ADC / payload tags

Explore tab

  • sparse SAR, cross-scaffold pairs, expansion leaves
  • weakly linked nodes and approved-tag neighborhood gaps
  • relation graph + stored Morgan fingerprints
  • category filters

Exploration-priority bins

  • High (X-S): top ~12%
  • Medium (X-A/X-B): ~12–62%
  • Low (X-C): remainder (pools under 5 compounds stay Medium)

Internal exploration metric based on catalog and structural signals. Not a prediction of efficacy, safety, potency, or development success.

Full methodology

Assessed: 9/11/2026, 5:36:19 AM · 5 min cache · All families