Fig. 5. Improvements in scoring, structure sampling, and binding orientation sampling should increase design binding affinity and selectivity.

a Conformational sampling of des1.1.0 (Source Data are provided as a Source Data file). Inset: Overlay of the crystal structure (gray) and the best scoring model (green). The CA atom of the SHA anchor is shown as sphere for reference. b Overlay of predicted binding orientation from our large-scale parallel docking (dark gray) with crystal structure (light gray) of des1.1.0 shows that docking can accurately predict orientation of key residues at the interface. c Despite improvement of the IC₅₀ values over the original SHA anchor (pink square), most designs follow the same trend as SHA; binding slightly more tightly to HDAC6 over HDAC2. Different colors show results of designs from different methods (blue = method 1, orange = method 2, green = method 3, purple = method 4). TSA (Trichostatin A, red diamond), a pan-HDAC inhibitor, is shown as a control. d Comparison of computational conformational sampling for a structured (gray) and a functional (purple) macrocycle shows a much deeper energy gap for structured macrocycle compared to the functional macrocycle (Source Data are provided as a Source Data file).