Computational structure-activity study directs synthesis of novel antitumor enkephalin analogs - QSAR dataset

This dataset accompanies the publication "<strong>Computational structure–activity study directs synthesis of novel antitumor enkephalin analogs</strong>" by Gredičak, Supek et al. (2010) Amino Acids. doi:10.1007/s00726-009-0329-5 The training data includes 22 chemical compounds with known biological activity, obtained via principal components analysis (PCA) of many measurements of cytotoxicity against multiple cancer cell lines. The test data is a virtual library of 429 compounds, for which biological activity was predicted using Support Vector Regression. Four best candidates were synthesized and tested for activity against tumor cell lines (see publication for details). All compounds are peptides which may include unnatural, adamantane-containing amino acids. The 1666 descriptors were computed using the E-Dragon online service.

本数据集配套Gredičak、Supek等人2010年发表于《氨基酸》（Amino Acids）期刊的论文《Computational structure–activity study directs synthesis of novel antitumor enkephalin analogs》，其DOI为10.1007/s00726-009-0329-5。训练集包含22种具备明确生物活性的化合物，这些化合物由针对多种癌细胞系的细胞毒性多组测量数据经主成分分析（Principal Components Analysis）计算得到。测试集为包含429种化合物的虚拟化合物库，其生物活性通过支持向量回归（Support Vector Regression）完成预测。研究人员已合成4种最优候选化合物，并针对肿瘤细胞系开展了活性测试，详细信息请参阅该发表论文。所有受试化合物均为肽类，可包含含金刚烷基的非天然氨基酸。本数据集的1666个分子描述符均通过E-Dragon在线服务计算得到。