Supplementary files for the article "Inverse Similarity and Reliable Negative Samples for Drug Side-effect Prediction"

Supplementary file 1: The supplementary figures for this work. Figure S1: Scatter plots of F1-scores for different classifiers using ChemTar similarity on balanced and unbalanced dataset. Figure S2: Scatter plots of F1-scores for ComNegative and ChemTarRandom using different classifiers. Supplementary file 2: The supplementary tables for this work. Table S1: List of 917 drugs studied in this work. Drug names, drug target proteins, substituents of drugs, ATC codes and the SMILES strings are included as well. Table S2: List of 500 side-effect terms studied in this work. Table S3: F1-scores of different similarity measurements and different similarity integration methods using KNN. Table S4: F1-scores of different classifiers using the comprehensive similarity on balanced and unbalanced training sets. Table S5: F1-scores of different classifiers using the proposed negative sample selection method and random sample selection method.补充文件1：本研究的补充附图。图S1：在平衡与非平衡数据集上采用ChemTar相似度的不同分类器的F1分数散点图。图S2：ComNegative与ChemTarRandom在不同分类器下的F1分数散点图。补充文件2：本研究的补充附表。表S1：本研究纳入的917种药物列表，包含药物名称、药物靶蛋白、药物取代基、ATC编码（ATC codes）以及SMILES字符串（SMILES strings）。表S2：本研究纳入的500种副作用术语列表。表S3：采用K近邻（KNN）时，不同相似度度量方法与相似度整合方法的F1分数。表S4：在平衡与非平衡训练集上采用综合相似度时，不同分类器的F1分数。表S5：分别使用本文提出的负样本选择方法与随机样本选择方法时，不同分类器的F1分数。