De Novo Design of Hydrogen-Bonding Networks for Deterministic Bio-Nano Recognition

Abstract Achieving deterministic control over molecular recognition at high-curvature bio-nano interfaces remains a fundamental challenge. For two decades, DNA-mediated single-wall carbon nanotube (SWCNT) sorting has relied on empirical screening, leaving the underlying recognition code largely undeciphered. Here, we report a de novo design framework that rationally programs DNA sequences via a generalized hydrogen-bonding network (HBN) model. Navigating the vast sequence space via an automated HBN-driven compiler, we experimentally evaluated a library of 150 designed sequences. This strategy achieved an extraordinary 91.3% success rate in mediating chirality-specific sorting, facilitating the high-purity isolation of 21 distinct (n, m) SWCNT species. Notably, this framework enabled the capture of three rare quasi-metallic nanotubes─(8,2), (9,3), and (10,1)─which were previously inaccessible via conventional screening. Mechanistically, the helical periodicity (N) serves as a universal tuning knob to coordinately modulate nanotube diameter and interfacial DNA pitch. This structural programmability enables the “bespoke” engineering of DNA-SWCNT sensors with tailored molecular discrimination and electrochemical resilience. This work transitions the field from labor-intensive, trial-and-error sequence screening to a deterministic, rational design paradigm.