Phase 8h-lite #5 · Tanimoto similarity scan: lead vs known cochlear ototoxins

Goal. Quick chemical-class sanity check: does the lead share Morgan-FP motifs with any known ototoxin class?

Method. RDKit Morgan FP (r=2, 2048 bits), Tanimoto similarity, lead v5.2__aq3__adamantyl__CONHOH__-Cl vs 12-compound ototoxin panel covering:

  • aminoglycosides (gentamicin, kanamycin, amikacin, neomycin)
  • loop diuretics (furosemide, ethacrynic acid)
  • platinum (cisplatin, carboplatin)
  • salicylates (aspirin)
  • macrolide (erythromycin)
  • quinoline (quinine)
  • glycopeptide (vancomycin)

Source: Schacht 2008 Auditory Trauma, Protection and Repair — RAG-indexed in BookLibrary.

Script: tanimoto_ototoxins.py. Run time <1 s.

Result

Ototoxin classCompoundTanimoto
Aminoglycosidegentamicin C1parse-fail
Aminoglycosidekanamycin A0.061
Aminoglycosideamikacin0.086
Aminoglycosideneomycin B0.071
Loop diureticfurosemide0.120
Loop diureticethacrynic acid0.103
Platinumcisplatin0.000
Platinumcarboplatinparse-fail
Salicylateaspirin0.127
Macrolideerythromycin core0.096
Quinolinequinine0.101
Glycopeptidevancomycin coreparse-fail

Max similarity: 0.127 (aspirin), well below the 0.40 flag threshold.

Interpretation

  • The lead carries no Morgan-FP motif in common with any major class of cochlear ototoxin.
  • The closest match is aspirin (salicylate, 0.127), driven by the shared aromatic + carboxamide-like polar headgroup; this is far from the salicylate motif that drives reversible salicylate ototoxicity.
  • The aminoglycoside panel max is 0.086 — the lead’s adamantyl + aza-quinoline scaffold has no overlap with the polyamine-sugar scaffold class that drives most clinical permanent ototoxicity.
  • Three SMILES failed to parse (gentamicin C1, carboplatin, vancomycin core fragment due to ring-numbering errors in the inputs); the ones that did parse cover every major mechanism class.

Caveat

Tanimoto on Morgan FP captures shared substructure, not ototoxicity per se. Many ototoxic drugs cross-react via mechanism-specific transporters (cisplatin via CTR1, aminoglycosides via TRPML/MET) or generate ROS — these mechanisms are not predicted from chemical similarity. Wet-lab cytotoxicity panel against HEI-OC1/UB-OC1 cell lines remains required (Phase 8 SOP).

Connections