Predicting Small\nMolecule Binding Nucleotides in RNA\nStructures Using RNA Surface Topography

RNA small molecule interactions play\na crucial role in\ndrug discovery\nand inhibitor design. Identifying RNA small molecule binding nucleotides\nis essential and requires methods that exhibit a high predictive ability\nto facilitate drug discovery and inhibitor design. Existing methods\ncan predict the binding nucleotides of simple RNA structures, but\nit is hard to predict binding nucleotides in complex RNA structures\nwith junctions. To address this limitation, we developed a new deep\nlearning model based on spatial correlation, ZHmolReSTasite, which\ncan accurately predict binding nucleotides of small and large RNA\nwith junctions. We utilize RNA surface topography to consider the\nspatial correlation, characterizing nucleotides from sequence and\ntertiary structures to learn a high-level representation. Our method\noutperforms existing methods for benchmark test sets composed of simple\nRNA structures, achieving precision values of 72.9% on TE18 and 76.7%\non RB9 test sets. For a challenging test set composed of RNA structures\nwith junctions, our method outperforms the second best method by 11.6%\nin precision. Moreover, ZHmolReSTasite demonstrates robustness regarding\nthe predicted RNA structures. In summary, ZHmolReSTasite successfully\nincorporates spatial correlation, outperforms previous methods on\nsmall and large RNA structures using RNA surface topography, and can\nprovide valuable insights into RNA small molecule prediction and accelerate\nRNA inhibitor design.