Fast and Reliable NMR-Based Fragment Scoring for Drug Discovery.

Area of Science:

• Biochemistry and Structural Biology
• Medicinal Chemistry
• Biophysics

Background:

• Fragment-Based Drug Discovery (FBDD) identifies novel therapeutics by screening molecular fragments against protein targets.
• Nuclear Magnetic Resonance (NMR) spectroscopy is crucial for detecting weak protein-ligand interactions in FBDD.
• Traditional NMR methods for measuring binding affinities (K_D) are time-consuming and labor-intensive.

Purpose of the Study:

• To develop an innovative and efficient method for fragment scoring in FBDD.
• To accelerate the quantification of fragment binding affinities (K_D) using NMR.
• To integrate machine learning with NMR for faster hit-to-lead compound transition.

Main Methods:

• Developed a SHARPER NMR technique for sensitive detection of weak protein-ligand interactions.
• Implemented a machine learning model to rank fragment affinities using minimal SHARPER titration data points.
• Integrated SHARPER NMR with machine learning into the "ML-boosted ¹H LB SHARPER NMR" method.

Main Results:

• The SHARPER NMR technique significantly reduces data acquisition times for K_D measurements.
• The ML-boosted method accurately ranks fragment affinities using only two titration points.
• Up to 144 ligand K_D values were determined in a single day, a substantial improvement over traditional methods.

Conclusions:

• The ML-boosted ¹H LB SHARPER NMR method offers significant time savings in FBDD.
• This approach accelerates the evaluation of fragment binding, enabling faster decision-making in early drug discovery.
• The methodology streamlines the transition from initial fragment hits to lead compounds.