转移学习框架用于PTP1B抑制剂活性预测:通过Web平台实施自然和非自然产品的差异模型
Zixiao Wang1, Lili Sun2, Anqi Ren3
1Department of Pharmacy, Honghui Hospital, Xi' an Jiaotong University, Xi'an, 710054, China. zixiaowang1112@foxmail.com.
Molecular diversity
|November 20, 2025
概括
这项研究开发了定制的深度学习模型,用于预测蛋白氨酸酸酶1B (PTP1B) 抑制剂,通过区分天然和非天然产品来改善糖尿病,肥胖和癌症的药物发现.
科学领域:
- 药用化学 医学化学
- 计算化学的计算化学
- 药理学 药理学是指药理学的学科.
背景情况:
- 蛋白氨酸酸酶1B (PTP1B) 是代谢疾病和癌症的关键治疗点.
- 开发具有选择性和生物可用性的PTP1B抑制剂存在重大挑战.
- 现有的深度学习 (DL) 模型往往无法解释自然产品 (NP) 和非自然产品 (NNP) 之间的化学空间差异.
研究的目的:
- 创建专门的DL模型用于PTP1B抑制剂活性预测,区分NP和NNP.
- 通过转移学习 (TL) 提高NP抑制剂的预测性能.
- 为PTP1B抑制剂查提供一个用户友好的网络平台.
主要方法:
- 将化学数据集分成NP和NNP子集,用于模型培训.
- 员工转移学习 (TL) 通过对NNP的预培训模型和对NP的微调.
- 评估了三个DL架构:注意力FP (AFP),多层感知器 (MLP) 与ECFP4,以及基于SMILES的变压器 (P10M).
- 使用五倍交叉验证进行优化和绩效评估.
主要成果:
- 注意力FP (AFP) 在图形神经网络中表现优越.
- 使用扩展连接指纹4 (ECFP4) 的MLP模型显示出强的结果.
- 基于SMILES的变压器与PubChem10M_SMILES_BPE_450k (P10M) 在他们的类别中脱而出.
- 转移学习的NP模型始终优于其非TL对应模型.
- TL显著改善了NP活动预测和模型概括.
结论:
- 使用定制的DL模型和TL的新策略增强了PTP1B抑制剂查.
- 开发的模型和网络平台为针对PTP1B的药物发现提供了有价值的工具.
- 区分NP和NNP化学空间对于基于DL的有效药物设计至关重要.
相关概念视频
Pharmacogenetics of Phase II Enzymes: N-acetyltransferase, Thiopurine S-methyltransferase, UDP-glucuronosyltransferase
Phase II biotransformation reactions are essential for detoxifying and eliminating xenobiotics, including many pharmaceutical compounds. These reactions typically involve conjugation, the covalent attachment of polar endogenous groups such as glucuronic acid, sulfate, methyl, or acetyl moieties to functional groups introduced during Phase I metabolism. The resulting conjugates are more water-soluble, enabling efficient renal or biliary excretion.The major classes of Phase II enzymes include...
Pharmacogenomics: Identification of New Drug Targets
Advances in genomics have profoundly influenced drug discovery by increasing both the speed and accuracy of pharmaceutical development. Pharmacogenomics, which examines how genetic variation influences drug response, facilitates the identification of novel therapeutic targets and enables patient stratification for personalized treatment. These strategies contribute to improved drug efficacy, minimized adverse effects, and more efficient clinical trial design.Mapping genetic differences...


