本文通过比较不同分析流程（rMATS、SplAdder、SUPPA2）和建库方法（PolyA和RiboZ）在选择性剪接事件分析中的性能，寻找识别选择性剪接的最佳实践方法。研究采用SNR、检出数、交并比和皮尔森相关系数等指标，对分析结果进行了信噪比、检出一致性和定量一致性的评估。结果表明，PolyA建库方法在数据质量和性能上优于RiboZ。SplAdder分析流程的选择性剪接结果信噪比与定量一致性更优，在选择性剪接事件识别时表现出较高的稳定性和可靠性。而SUPPA2分析流程在检出一致性的表现上更佳，能够识别到更广泛的选择性剪接事件。此外，通过TPM和Ratio处理校正批次效应后，SUPPA2的性能得到了显著提升。本研究为选择性剪接事件的检测与分析提供了有益的思路，并为未来的抗癌药物研发与精准医疗研究提供了新的参考。

This study compares the performance of various analysis pipelines (rMATS, SplAdder, SUPPA2) and library preparation methods (PolyA, RiboZ) in detecting alternative splicing events. Using SNR, detection numbers, IoU, and Pearson correlation coefficient, we evaluated the results' signal-to-noise ratio, detection consistency, and quantitative consistency. The PolyA method excelled in data quality and performance, while SplAdder demonstrated high SNR and quantitative consistency. SUPPA2 showed better detection consistency and improved performance after TPM and Ratio correction. This study offers insights for alternative splicing detection and analysis, supporting anticancer drug development and precision medicine research.