π-USEA:一种同步推断泛素连接酶与去泛素化酶活性的新工具OA

OBJECTIVE To develop a computational tool designed for the synchronous and high-sensitivity inference of ubiquitin ligase(E3)and deubiquitinase(DUB)activities.METHODS Large-scale enzyme-substrate interaction reference sets were constructed by integrating experimentally vali-dated and computationally predicted data.A dynamic threshold optimization strategy was adopted to filter high-confidence interactions.Furthermore,a directional weighted Z-score model was designed to assign weights based on interaction confidence,which enabled unified modeling of the opposing regu-latory roles of E3s and DUBs within a single statistical framework.The accuracy of E3 activity inference was validated using the speckle-type pox virus and zinc finger protein(SPOP)-encoding gene-mutant prostate cancer dataset,while the DUB inference and biological interpretability were determined using the lipopolysaccharide(LPS)-stimulated macrophage time-series dataset(0.25,2,and 4 h).RESULTS The integrated prediction set of π-USEA covered 781 human E3s,and for the first time incorporated the DUB interaction network.In the SPOP dataset,π-USEA accurately recapitulated the patterns of wild-type(WT)activation and mutant dominant-negative inhibition of various SPOP mutants(MTs)(WT vs Control:Z=6.31,P=2.76×10-10;MTs vs WT:Z<-7,P≤1×10-12),achieving a 5-9 orders of magnitude improvement in statistical significance(P value)over UbE3-APA.In the LPS time-series analysis,π-USEA correctly captured the rapid transient activation of tumor necrosis factor α-induced protein 3(TNFAIP3)and sustained downregulation of ubiquitin specific peptidase 7(USP7),and substrate enrichment analysis revealed a temporal shift in regulatory focus from early metabolic repro-gramming to late translational regulation.CONCLUSION By integrating bidirectional interaction data with a dynamic weighting strategy,π-USEA significantly enhances the accuracy and statistical sensitivity of E3 and DUB activity predictions.