Ligand Efficiency Metrics — Catalog

P0 reference — verbatim formulas and recommended ranges from 2014-schneider-de-novo-molecular-design-book §6.4.1 (Mazanetz, Law, Whittaker, pp. ~165–167) and §1.4 (Schneider & Baringhaus, Eq. 1.8). The metric stack used to prioritize fragments for elaboration and to monitor MPO progress through hit-to-lead.

LE itself: cited from Hopkins, Groom, Alex, Drug Discov. Today 9 (2004) 430–431 [50 / 29 in book].

Metric definitions (verbatim formulas)

1. Ligand Efficiency (LE) — Eq. 1.8 / §6.4.1 base

LE = ΔG / N = −RT ln(K_D) / N (kcal/mol per heavy atom)

Schneider 2014 §5.3.3 alternative form using IC₅₀:

LE ≈ −0.59 (ln IC₅₀) / HAC (at 298 K)

where HAC = heavy-atom count (non-hydrogen atoms), R = 1.987 × 10⁻³ kcal/(mol·K), T = absolute temperature.

2. Fit Quality (FQ) — Reynolds et al. [226]

FQ = LE / LE_Scale

LE_Scale = −0.064 + 0.873 · exp(−0.026 · HAC)

— Schneider 2014 §6.4.1 (1).

3. Size-Independent Ligand Efficiency (SILE) — Nissink [227]

SILE = affinity / HAC^0.3

where affinity is pK_i or pIC_50.

— Schneider 2014 §6.4.1 (2).

4. Group Efficiency (GE) — Verlinde / Ferenczy [228]

GE = −ΔΔG_b / ΔHAC

The binding-free-energy contribution of a functional group A→B addition, per heavy atom added.

— Schneider 2014 §6.4.1 (3). Caveat: assumes group-additivity; see Fragment Additivity Assumption and Superadditivity.

5. Ligand Lipophilicity Efficiency (LLE / LiPE) — Leeson [2]

LLE = −pIC₅₀ (or pK_i) − cLogP (or LogD)

— Schneider 2014 §6.4.1 (4).

6. Ligand-Efficiency-Dependent Lipophilicity (LELP) — Keserű & Makara [229]

LELP = log P / LE

— Schneider 2014 §6.4.1 (5).

7. Ligand Lipophilicity Efficiency AT (LLE_AT) — Astex / Mortenson & Murray [230]

LLE_AT = (0.11 · ln(10) · RT · (Log P − Log(activity))) / HAC (kcal/mol per heavy atom)

where activity is K_D or IC_50.

— Schneider 2014 §6.4.1 (6). The Astex-derived size-aware LLE; recommended over plain LLE for fragments.

8. Kinetic Efficiency (KE) [233]

KE = t_½ / (0.693 · HAC)

where t_½ is target-residence-time half-life. Late-stage-only metric.

— Schneider 2014 §6.4.1.

MetricRangeInterpretation
LE≥0.3 kcal/mol per HAtypical fragment-elaboration target (drug at MW 500 Da, IC₅₀ = 10 nM, ~38 HA → LE = 0.3)
LE≈ −1.5 kcal/mol per HA at HAC ≥ 15binding-energy contribution plateau (Kuntz max-affinity ceiling)
Per-atom optimization gain0.29 kcal/mol per non-H atommean affinity addition per added heavy atom during MPO [146, 222]
FQ≈ 1.0optimal binding
FQ< 0.6poor binding
SILE> 2.5optimization target
LLE / LiPE5–7+aim during MPO; ensures non-promiscuous final compound
LELP< 16.5”Lipinski-compliant” ceiling
LELP (lead range)−10 to +10optimize toward 0 with improving potency

Method-choice survey (from §6.4.1 closing)

“A survey conducted on the blog web site ‘Practical Fragments’ has highlighted LE followed by LLE as being the two most popular metrics.”

— Schneider 2014 §6.4.1 [231].

The Mazanetz lab’s recommendation: use LE for fragment selection; track LE + LLE during optimization. LLE_AT is the preferred lipophilicity-aware metric at the fragment stage because it incorporates HAC.

How to use in STRC

  • h01 phase 3b fragment-pocket-fit scoring: record LE per fragment in the ranked output; gate at LE ≥ 0.3 kcal/mol per HA.
  • h01 v4 fragment-grow: track GE for each added group; if a substituent’s GE < 0.2 kcal/mol per HA, the addition is wasted size — revert.
  • h01 phase 4 docking pose ranking: convert Vina ΔG (kcal/mol) and ligand HAC to LE; use LE-rank as a parallel sort key alongside raw Vina score (Vina-score alone biases toward larger ligands per §6.3.9 [185]).
  • h01 phase 7+ pharmaceutical filter: require LLE_AT > 0.3 kcal/mol per HA at the lead stage as a defense against the Leeson “MW + log P creep” attrition pattern (Leeson 2011 [83]).
  • Citation pattern: “ligand-efficient fragment scoring per Schneider 2014 §6.4.1 (Mazanetz et al.); LE original definition Hopkins, Groom, Alex 2004 Drug Discov. Today.”
  • Caveat (per §6.4.1 [233]): KE (kinetic efficiency) is currently “of limited application to fragments” — fragments have intrinsically fast on/off kinetics. Don’t apply at hit stage.

Connections