STRC Research

Druggability vs Ligandability Distinction

4 min read

Druggability vs Ligandability — the Distinction That Matters

P2 concept — distilled from 2014-schneider-de-novo-molecular-design-book §6.3.10 (Mazanetz, Law, Whittaker, p. ~163). A subtle but load-bearing distinction that is routinely conflated in the computational-chemistry literature, including in the h01 audit fixes already documented in pharmacochaperone.

Definitions

Ligandability: the property of a binding site that small molecules can bind to with detectable affinity. Detected by fragment screen, by computational hot-spot mapping (FTMap, GRID, MCSS, SILCS), or by exhaustive solvent-mapping (Ringe MSCS).

Druggability: the more demanding property that a binding site can yield a clinically usable drug. Adds requirements:

“a ligandable target may in fact have poor affinity for molecules with the ideal physicochemical (e.g., solubility, logD) and ADMET (e.g., no effects on hERG, Nav1.5, and CYPs) characteristics and required binding kinetics (on- and off-rates) necessary to become a drug.” — Schneider 2014 §6.3.10

For druggability, you also need:

  • A reachable potency window with drug-like physicochemistry (Lipinski rule-of-five compliant).
  • ADMET selectivity (no hERG / Nav1.5 / CYP3A4 promiscuity).
  • Achievable on/off-rate kinetics (residence time tuned to the disease).
  • Either family-wide undruggability (PPIs, generally hard) or precedent (Class A GPCRs, kinases — generally druggable).

Why this distinction is operationally critical

Schneider 2014 §6.3.10 quotes:

“Ligand interaction hot spots on the surface of proteins can be identified experimentally by Ringe’s MSCS method… These experimentally located binding hot spots correlate well with computational fragment mapping programs FTMap and HotspotsX.”

— but identifying a ligand hot-spot (i.e., demonstrating ligandability) is not the same as proving the target is druggable. The h01 phase 2/2b druggability score is — strictly — a ligandability estimator. Schneider 2014 makes the same point: many computational tools labeled “druggability assessment” actually predict ligandability.

Implications for STRC h01

h01 phase 2 / 2b “druggability” scores are ligandability scores

The current h01 phase 2 and 2b scripts in models/ compute composite scores derived from SiteMap / Halgren 2009 conceptual primitives (volume v_opt, hydrophobicity, depth) — see pharmacochaperone row “Druggability v_opt” and “Druggability composite weights”. These tell us:

  • ✅ The E1659A subpocket is ligandable (composite score 0.86 WT, gates passed at 0.70).
  • ❌ They do NOT tell us E1659A binders will be druggable in Misha’s clinical context (RWM penetration, OHC delivery, no off-target hERG/Nav1.5).

Action for h01 docstrings

Update docstrings to say “ligandability score” rather than “druggability score” where the distinction matters. The composite metric is a ligandability metric per the Schneider 2014 §6.3.10 framing. Druggability for h01 is established only when:

  1. Phase 5 / 5e MD ΔΔG ≤ rescue threshold AND
  2. RWM PK-envelope test passed (MW / logP / HBD ranges) AND
  3. Phase 7 panel hERG/Nav1.5 scaffolds-clear AND
  4. Phase 8 cross-target panel cleaner than ChEMBL background.

This compounds into the classical Class 1 target ideal of Schneider 2014 Fig. 6.5: both validated biologically AND has precedented druggability.

Schneider 2014 recommendation

“AstraZeneca have recommended the use of an initial fragment screen to determine the ligandability of new targets… The ligandability score reflects an assessment of hit rate, best affinities, and diversity obtained in fragment screening as follows:

  • high ligandability = high hit rate, best affinities <0.1 mM, larger diversity;
  • medium ligandability = intermediate hit rate, best affinities 0.1–1.0 mM, some diversity;
  • low ligandability = low hit rate, best affinities >1 mM, low diversity of hits.”

— Schneider 2014 §6.3.10 [216].

Application: h01 wet-lab ligandability gate

If h01 ever advances to wet-lab ligandability validation (e.g., before committing to a 1000+ compound virtual screen), apply the AstraZeneca rubric:

  • Test ~100 fragments at high concentration against E1659A STRC region by SPR or BLI.
  • Measure hit rate, best K_d, and chemotype diversity of hits.
  • Classify ligandability tier per the rubric above.
  • This output is much more informative than another in-silico “druggability” score.

Implications for other STRC hypotheses

  • h02 (piezoelectric): target is a small-molecule bioelectronic ligand of OHC mechanosensors. Ligandability completely unknown — no wet-lab fragment screen exists for this target class. Tier B is currently appropriate.
  • h05 (calcium oscillation): target involves voltage-gated channels. Druggability is high for this protein class (precedent in cardiac drugs) but Misha-fit is the limiter (per STRC Hypothesis Ranking).
  • h09 (hydrogel): hydrogel is not a ligand-binding hypothesis — ligandability/druggability framework does not apply. Replace with assembly fitness + delivery feasibility.
  • h26 (engineered homodimer): target is the protein–protein interface. PPI ligandability is “broad range from druggable to undruggable” per Schneider 2014 §6.3.10. h26’s strategy of engineering the interface rather than ligating it sidesteps the ligandability question entirely — a strategic advantage.

Connections

Graph View

Backlinks