基于分布感知与扩张卷积金字塔的SNO位点预测方法OA

S-nitrosylation(SNO)is a reversible post-translational modification of proteins that plays a crucial regulatory role in various physiological and pathological processes.To overcome the susceptibility of existing methods to false negative interfe-rence,cross-channel semantic redundancy,and challenges in semantic collaboration,the research integrates sample filtering with semantic modeling for SNO site prediction.This paper proposed a prediction model,SFEP-SNO,which integrated distri-bution awareness with semantic fusion.The model actively identified and removed false negative samples through a distribution-aware mechanism,thereby reducing their interference during training.In parallel,it employed an expanded convolutional py-ramid based on biologically driven feature encoding to achieve multi-scale semantic fusion and deep representation of key fea-tures.Experimental evaluations demonstrate that SFEP-SNO consistently improves accuracy,specificity,and AUC across mul-tiple independent datasets,confirming its capability to enhance discriminative performance in SNO site prediction and to facili-tate research on SNO-related mechanisms and disease diagnosis.