Fast and Reliable NMR-BasedFragment Scoring for DrugDiscovery

Fragment-Based Drug Discovery (FBDD) is a powerful strategy used in the development of new therapeutics. Molecular fragments are screened against a target protein, where interactions are typically characterized by a low affinity. Nuclear Magnetic Resonance (NMR) spectroscopy is well-suited to detect weak protein–ligand interactions and is therefore often used in FBDD. However, while NMR is very effective in initial screening, follow-up NMR experiments to measure binding affinities (i.e., <i>K</i>_D values) are labor-intensive and time-consuming. To address this challenge, we have developed an innovative SHARPER NMR fragment scoring technique. The high sensitivity of SHARPER NMR dramatically reduces the data acquisition times, allowing faster and more accurate quantification of fragment <i>K</i>_D values from ligand titration curves. To further accelerate fragment scoring, a machine learning model was developed that accurately ranks fragment affinities from only two SHARPER titration points. The resulting integrated method, termed “ML-boosted ¹H LB SHARPER NMR”, produced significant time savings; using a 600 MHz QCI cryoprobe, <i>K</i>_D values of up to 144 ligands in a day could be determined under our conditions, compared with only a handful achievable by traditional approaches. The proposed methodology will shorten the transition from hits to lead compounds, accelerating the drug discovery process by rapidly and reliably evaluating fragment binding, providing informed decision-making in the early stages of FBDD.