ADMET & In-Silico Drug Design

ADMET & In-Silico Drug Design

ADMET — Absorption, Distribution, Metabolism, Excretion, and Toxicity — encompasses the pharmacokinetic and safety properties that determine whether a drug candidate can become a medicine. Optimising ADMET early in drug discovery, using computational (in-silico) prediction, reduces late-stage attrition.

Absorption

Oral absorption requires dissolution in GI fluid, permeation across the intestinal epithelium, and survival of first-pass metabolism. Passive transcellular permeation correlates with lipophilicity (logD), molecular weight, and polar surface area (PSA). Active transport (OATP, P-gp) can supplement or limit absorption. In-silico tools — ADMET Predictor, SwissADME — estimate Caco-2 permeability and P-gp efflux ratios from 2D structure alone.

Distribution

Distribution is described by volume of distribution (Vd). High Vd indicates extensive tissue binding; low Vd suggests confinement to plasma. Plasma protein binding (PPB) — primarily to albumin and α1-acid glycoprotein — affects free drug fraction. CNS penetration requires passage across the blood-brain barrier (BBB); predictors include logP, MW < 400, PSA < 90 Ų, and absence of P-gp substrates.

Metabolism

Most drugs are metabolised in the liver, primarily by cytochrome P450 (CYP) enzymes. CYP3A4 handles ~50% of all drugs; CYP2D6, CYP2C9, and CYP2C19 handle many others. Phase I reactions (oxidation, reduction, hydrolysis) introduce or unmask polar groups. Phase II reactions (glucuronidation, sulfation, acetylation) attach hydrophilic moieties for excretion. In-silico tools predict CYP substrate and inhibitor status, flagging potential drug-drug interactions and metabolically labile sites (e.g., unprotected para positions on aromatic rings).

Excretion

Renal excretion removes water-soluble metabolites via glomerular filtration and tubular secretion. Hepatic excretion via bile handles higher-MW compounds. Terminal half-life (t½) determines dosing frequency; t½ = 0.693 × Vd / CL.

Toxicity Prediction

In-silico toxicity screening identifies structural alerts (SALERTs) — functional groups associated with known toxic mechanisms: reactive electrophiles (Michael acceptors, epoxides), hERG channel blockers (cationic amphiphiles), mutagenic aromatic amines, and PAINS (pan-assay interference compounds). DEREK Nexus and the OECD QSAR Toolbox apply rule-based expert systems alongside machine-learning classifiers trained on in-vitro and in-vivo toxicity datasets.

Virtual Screening and Docking

Molecular docking scores binding poses of compound libraries against a protein target crystal structure. High-throughput virtual screening of millions of compounds narrows a library to hundreds of candidates for synthesis. Scoring functions estimate binding free energy (ΔG_bind). Limitations include rigid-receptor approximations, solvation inaccuracies, and entropy neglect — experimental validation of top-ranked hits is always required.