In silico-driven protocol for hit-to-lead optimization: a case study on PDE9A inhibitors
Hiroyuki Ogawa1,2, Masateru Ohta3, Mitsunori Ikeguchi4,5
1Graduate School of Medical Life Science, Yokohama City University, 1-7-29 Suehiro-cho, Tsurumi-ku, Yokohama, 230-0045, Japan.
Journal of Computer-Aided Molecular Design
|December 18, 2025
Summary
This study introduces an in silico hit-to-lead (H2L) optimization protocol. It uses molecular generation, non-equilibrium switching (NES) for binding affinity, and machine learning (ML) for ADME properties to explore chemical space efficiently.
Area of Science:
- Computational chemistry
- Drug discovery
- Medicinal chemistry
Background:
- Hit-to-lead (H2L) optimization is crucial for small-molecule drug discovery.
- Traditional H2L methods limit chemical space exploration due to iterative synthesis and evaluation.
- In silico approaches offer efficient exploration of vast chemical spaces via virtual compound generation and computational evaluation.
Purpose of the Study:
- To develop and validate an in silico-driven H2L protocol for efficient chemical space exploration.
- To assess the accuracy and utility of the non-equilibrium switching (NES) method for binding free energy calculations in H2L.
- To integrate molecular generation, NES-based affinity prediction, and machine learning (ML) for ADME property evaluation.
Main Methods:
- Developed an integrated in silico H2L protocol combining molecular generation, NES for relative binding free energy calculations, and ML for ADME property prediction (solubility, metabolic stability, permeability).
- Applied the protocol to a phosphodiesterase 9A inhibitor model system, starting from a high-throughput screening hit.
- Conducted large-scale exploration of substituent space at two key positions.
Main Results:
- The in silico protocol successfully identified compounds with high predicted binding affinity and favorable ADME properties.
- The lead compound previously reported in the literature was identified among the top-ranked candidates.
- NES method demonstrated effectiveness in large-scale substituent space exploration for H2L optimization.
Conclusions:
- An in silico H2L protocol integrating large-scale molecular generation, high-accuracy NES affinity prediction, and ML-based ADME prediction enables broader exploration of substituent space.
- This approach accelerates the identification of promising drug candidates with desirable properties.
- The study validates the utility of NES for precise binding free energy calculations in drug discovery workflows.
