Computer-Aided Drug Design (1): Predicting Drug Potency

Free Energy Calculation Can Predict Drug Potency Accurately

In drug discovery, the question “Is this compound effective?” can be answered with experiments and calculations. This article explains what drug potency means, how computational chemistry predicts it, and why free energy is key.

1. What is Drug Potency?

Drug potency measures how well a compound works on a target, usually shown by the concentration needed for an effect. Higher potency means a lower dose is required.

Common ways to measure potency include:

• IC₅₀: The concentration needed to block 50% of the target’s activity.

• EC₅₀: The concentration required for 50% of the maximum effect.

• K_d / K_i: Constants showing how strongly a compound binds to the target.

These are measured through experiments like enzyme tests or binding assays.

2. How Does Computational Chemistry Predict Binding Strength?

Testing compounds in a lab is expensive, so computational methods predict how well a molecule binds to a target.

Method 1: Single Conformation Energy Estimate
Calculates binding energy for one docking pose. It’s fast and cheap but ignores flexibility, so it’s less accurate.

Method 2: Free Energy Calculations
Uses statistical thermodynamics to account for multiple conformations and motion. More accurate methods include:

• Molecular Dynamics (MD) with MM/PBSA or MM/GBSA.

• Free Energy Perturbation (FEP) for high precision.

Method 3: AI Predictions
Uses trained data to predict IC₅₀ quickly, but is limited by data and less explainable.

The most accurate approach calculates binding free energy (ΔG_bind), the energy released when a molecule binds to a protein. Lower energy means stronger binding.

The formula for free energy is:

\(\Delta G_{\text{bind}} = R T \cdot \ln K_i\)

Where K_i is the binding constant:

\(K_i = \frac{[L][P]}{[PL]}\)

[L] is the drug concentration, [P] is the protein concentration, and [PL] is the drug-protein complex concentration. Lower K_i means stronger binding and lower free energy.

Free energy is calculated using Molecular Dynamics (MD) or Monte Carlo sampling.

4. Linking Free Energy to Potency (IC₅₀)

IC₅₀ is related to K_i. Free energy can estimate IC₅₀, but it’s not perfect because IC₅₀ is measured in complex conditions.

The Cheng-Prusoff equation connects them:

\(K_i = \frac{IC_{50}}{1 + \frac{[S]}{K_m}}\)

Where [S] is substrate concentration and K_m is the substrate’s binding constant. When [S] = K_m, K_i equals IC₅₀.

Comparing two compounds’ free energy difference (ΔΔG) predicts their IC₅₀ ratio:

\(\begin{align*} \Delta \Delta G^{A \rightarrow B} &= \Delta G_{\text{bind}}^B - \Delta G_{\text{bind}}^A \\ &= RT \ln \left( \frac{K_i^B}{K_i^A} \right) \\ &\approx RT \ln \left( \frac{IC_{50}^B}{IC_{50}^A} \right) \end{align*}\)

This shows how structural changes affect potency.

5. Example: Designing SARS-CoV-2 M^pro Inhibitors

In a 2021 study by William L. Jorgensen, researchers designed non-covalent inhibitors for the SARS-CoV-2 main protease (M^pro). They used Free Energy Perturbation (FEP) to calculate ΔΔG for structural changes and predict which changes improved potency. The table shows results for compounds 4, 5, 6, and 9:

The trend matches well: larger negative ΔΔG (like for compounds 5 and 6) means lower IC₅₀ (higher potency). Compound 9 has a smaller ΔΔG, so its IC₅₀ is higher (less potent).

6. Best Use of Free Energy: Optimizing Lead Compounds

Imagine a lab finds compound A with a decent IC₅₀, but it’s not good enough. The team suggests 100 possible tweaks, but testing them all takes months. Computational design steps in: calculate ΔΔG for each tweak compared to compound A, rank them, and prioritize the best ones for synthesis.

This approach worked in developing Nirmatrelvir (Paxlovid). Pfizer used calculations to screen candidates and found a key five-membered ring structure.

Summary and Discussion

• Potency is just one part of drug development. Toxicity, bioavailability, and other factors matter too. Good potency doesn’t guarantee clinical success.

• ΔΔG predicts IC₅₀ changes and guides drug design. Physical models are more reliable and explainable than AI, but free energy calculations are sensitive to details like force fields.

• While others test many compounds, calculations narrow down candidates, speeding up discovery. Computational tools (CADD or AIDD) help small molecules or peptides lead the race.

Our computational chemistry team, with over 10 years of experience, works closely with experimental scientists to make high-quality calculations easy. Want to discuss drug development needs? Feel free to message or comment below!