ADMET Predictions

ADMET scoring predicts key pharmacokinetic and physicochemical properties: Absorption, Distribution, Metabolism, Excretion, and Toxicity.

Available Predictions

Property	Method	Output	Interpretation
Synthetic Accessibility	SAscore	1-10 (1=easy, 10=hard)	Ease of synthesis
Aqueous Solubility	ESOL (Delaney)	LogS, mg/mL, classification	Water solubility
Complexity	Fsp3, stereocenters, rings, Bertz CT	Various metrics	Molecular complexity
CNS MPO	Multi-parameter optimization	Score 0-6	CNS penetration likelihood
Bioavailability	TPSA, rotatable bonds, Lipinski	Predictions	Oral absorption, CNS flags
Pfizer 3/75 Rule	LogP < 3, TPSA > 75	Pass/fail	Toxicity risk reduction
GSK 4/400 Rule	MW ≤ 400, LogP ≤ 4	Pass/fail	Lead-like properties
Golden Triangle	MW vs LogD plot	In/out	Optimal drug-like space

Synthetic Accessibility (SAscore)

Estimates how difficult a molecule is to synthesize:

Score	Classification	Interpretation
1-3	Easy	Simple synthesis, few steps
4-6	Moderate	Standard synthetic methods
7-9	Difficult	Complex, many steps
10	Very Difficult	Extremely challenging

Based on fragment contributions and complexity penalties.

Aqueous Solubility (ESOL)

Predicts water solubility using the Delaney ESOL linear regression model.

ESOL Equation

LogS = 0.16 − 0.63 × LogP − 0.0062 × MW + 0.066 × RotBonds − 0.74 × AP

Where:

LogS = log₁₀(aqueous solubility in mol/L)
LogP = Wildman-Crippen LogP
MW = molecular weight (Da)
RotBonds = rotatable bond count
AP = aromatic proportion (aromatic atoms / heavy atoms)

Conversion: solubility_mg_mL = 10^LogS × MW / 1000

Classification

LogS	Category	Approx. mg/mL
≥ −1	Highly soluble	> 100
−1 to −3	Soluble	1–100
−3 to −4	Moderately soluble	0.1–1
−4 to −5	Poorly soluble	< 0.1
< −5	Insoluble	Very low

Good aqueous solubility (LogS > −4) is favorable for oral drugs.

Reference: Delaney (2004). ESOL: Estimating aqueous solubility directly from molecular structure. JCICS, 44(3), 1000–1005.

Molecular Complexity

Multiple complexity metrics:

Fsp3 (Fraction sp3 carbons):

Higher Fsp3 = more saturated, better 3D character
< 0.25: Flat/aromatic
0.25–0.42: Moderate 3D
> 0.42: Good 3D character (target for drug-likeness)

Reference (Fsp3): Lovering et al. (2009). Escape from flatland. Journal of Medicinal Chemistry, 52(21), 6752–6756.

Stereocenters:

Count of chiral centers
More stereocenters = higher synthetic complexity

Ring systems:

Total rings and aromatic rings
Complex scaffolds have many rings

Bertz CT (Complexity index):

Higher values = more complex
Accounts for branching and symmetry

CNS MPO (Multiparameter Optimization)

Pfizer's Central Nervous System Multiparameter Optimization score predicts CNS penetration likelihood on a 0–6 scale.

Component Scoring

Each component scores 0–1, and the total is the sum of all 6 components:

Parameter	Score = 1.0	Linear decrease	Score = 0
MW	≤ 360 Da	360–500 Da	> 500 Da
LogP	≤ 3.0	3.0–5.0	> 5.0
TPSA	≤ 40 A²	40–90 A²	> 90 A²
HBD	0	1–3 (−0.25 per donor)	> 3
LogD	0.5 (estimated)	—	—
pKa	0.5 (estimated)	—	—

Placeholder Values

LogD and pKa components use placeholder values (0.5 each) as these require experimental data or more complex models to calculate accurately.

Interpretation

CNS MPO Score	Interpretation
≥ 5	Excellent CNS penetration
4–5	Good CNS penetration
3–4	Moderate
< 3	Poor CNS penetration

Reference: Wager et al. (2010). Moving beyond rules: the development of a CNS MPO approach. ACS Chemical Neuroscience, 1(6), 435–449.

Bioavailability Predictions

Oral Absorption:

oral_absorption_likely = lipinski_ok AND veber_ok

Where lipinski_ok = MW ≤ 500, LogP ≤ 5, HBD ≤ 5, HBA ≤ 10 and veber_ok = rotatable bonds ≤ 10, TPSA ≤ 140.

CNS Penetration:

cns_penetration_likely = TPSA ≤ 90 AND MW ≤ 450 AND HBD ≤ 3 AND 1 ≤ LogP ≤ 4

Pfizer 3/75 Rule

A toxicity risk indicator:

at_risk = LogP > 3 AND TPSA < 75

Compounds meeting both criteria have statistically higher rates of toxicity and off-target promiscuity.

Reference: Hughes et al. (2008). Physiochemical drug properties associated with in vivo toxicological outcomes. Bioorganic & Medicinal Chemistry Letters, 18(17), 4872–4875.

GSK 4/400 Rule

GlaxoSmithKline's compound quality guideline:

favorable = MW ≤ 400 AND LogP ≤ 4

Provides room for optimization while maintaining favorable ADMET properties.

Reference: Gleeson (2008). Generation of a set of simple, interpretable ADMET rules of thumb. Journal of Medicinal Chemistry, 51(4), 817–834.

Golden Triangle

Abbott's optimal property space for balanced permeability and metabolic stability:

in_triangle = 200 ≤ MW ≤ 450 AND −0.5 ≤ LogP ≤ 5

Molecules within this space tend to have favorable permeability-metabolism balance.

Reference: Johnson et al. (2009). Using the Golden Triangle to optimize clearance and oral absorption. Bioorganic & Medicinal Chemistry Letters, 19(17), 5560–5564.

API Usage

curl -X POST http://localhost:8001/api/v1/score \
  -H "Content-Type: application/json" \
  -d '{
    "molecule": "CC(=O)Oc1ccccc1C(=O)O",
    "include": ["admet"]
  }'

Response:

{
  "admet": {
    "synthetic_accessibility": {
      "score": 1.5,
      "classification": "Easy",
      "interpretation": "Very easy to synthesize"
    },
    "solubility": {
      "log_s": -2.1,
      "solubility_mg_ml": 1.43,
      "classification": "Soluble",
      "interpretation": "Good aqueous solubility"
    },
    "complexity": {
      "fsp3": 0.11,
      "num_stereocenters": 0,
      "num_rings": 1,
      "num_aromatic_rings": 1,
      "bertz_ct": 245.2,
      "classification": "Low complexity"
    },
    "cns_mpo": {
      "score": 3.8,
      "cns_penetrant": false
    },
    "bioavailability": {
      "oral_absorption_likely": true,
      "cns_penetration_likely": false
    },
    "pfizer_rule": {
      "passed": true,
      "logp": 1.19,
      "tpsa": 63.6
    },
    "gsk_rule": {
      "passed": true,
      "mw": 180.16,
      "logp": 1.19
    },
    "golden_triangle": {
      "in_golden_triangle": true,
      "mw": 180.16,
      "logd": 1.19
    }
  }
}

Interpretation Guidelines

Good ADMET Profile:

SAscore < 5 (easy to synthesize)
LogS > -4 (good solubility)
Fsp3 > 0.4 (sufficient saturation)
Oral absorption likely
Passes Pfizer and GSK rules
In golden triangle

Poor ADMET Profile:

SAscore > 7 (hard to synthesize)
LogS < -6 (poorly soluble)
Very complex (many stereocenters, rings)
CNS penetration when not desired
Fails multiple rules

Limitations

All predictions are computational estimates:

Based on training data (mostly drug-like molecules)
May not generalize to unusual scaffolds
Don't replace experimental measurements
Use for prioritization, not absolute decisions

Experimental Validation Required

ADMET predictions guide early decisions but must be validated experimentally for lead candidates.

Use Cases

Lead Optimization

Track ADMET during optimization:

Monitor solubility changes
Avoid increasing synthetic complexity
Maintain favorable bioavailability
Stay in golden triangle

Compound Prioritization

Rank compounds by ADMET profile:

SAscore < 5 AND
LogS > -4 AND
oral_absorption_likely = true AND
pfizer_rule_passed = true

Library Design

Design screening libraries with good ADMET:

Target SAscore < 5
Ensure solubility > -4
Maintain Fsp3 > 0.4
Stay in golden triangle

Next Steps

Drug-likeness - Lipinski, QED, Veber rules
Scoring Overview - All scoring systems
API Reference - Full scoring API

Available Predictions​

Synthetic Accessibility (SAscore)​

Aqueous Solubility (ESOL)​

ESOL Equation​

Classification​

Molecular Complexity​

CNS MPO (Multiparameter Optimization)​

Component Scoring​

Interpretation​

Bioavailability Predictions​

Pfizer 3/75 Rule​

GSK 4/400 Rule​

Golden Triangle​

API Usage​

Interpretation Guidelines​

Limitations​

Use Cases​

Lead Optimization​

Compound Prioritization​

Library Design​

Next Steps​