FuseLinker:利用LLM预先训练的文本嵌入和域名知识来增强生物医学知识图的基于GNN的链接预测
Yongkang Xiao1, Sinian Zhang2, Huixue Zhou1
1Institute for Health Informatics, University of Minnesota, Minneapolis, MN, USA; Division of Computational Health Sciences, Department of Surgery, University of Minnesota, Minneapolis, MN, USA.
Journal of biomedical informatics
|September 26, 2024
概括
生物医学知识图的新框架FuseLinker通过整合结构,文本和域知识来增强链接预测. 这种方法显示出药物重用和临床应用的巨大潜力.
科学领域:
- 生物医学信息学 生物医学信息学
- 图形神经网络的神经网络
- 知识表示 知识表示
背景情况:
- 生物医学知识图 (BKG) 对于理解复杂的生物关系至关重要.
- 现有的链接预测模型往往无法充分利用BKG内部的各种信息.
- 准确的链接预测对于诸如药物重用等应用至关重要.
研究的目的:
- 开发FuseLinker,这是一个用于BKG中的链接预测的新框架.
- 整合结构,文本和领域知识,以提高预测准确度.
- 为了评估FuseLinker在基于图表的药物重新定位中的实用性.
主要方法:
- FuseLinker使用了与图形神经网络 (GNN) 结合的文本和域知识嵌入.
- 使用大型语言模型 (LLM) 生成文本嵌入,使用Poincaré图形嵌入表示域知识.
- 该框架融合了这些嵌入式,以学习BKG结构表示,用于链接预测.
主要成果:
- 在四个公共BKG上,FuseLinker显著超过了传统的知识图嵌入和基于GNN的模型.
- 在数据集中实现了高的平均互惠等级 (MRR) 和ROC曲线下的面积 (AUROC).
- 案例研究证明了FuseLinker在产生药物重用假设方面的能力.
结论:
- FuseLinker是生物医学链接预测的有效框架,集成多种信息来源.
- 该模型显示了生物医学和临床任务中的实际应用的巨大潜力.
- 源代码和数据是公开可用的,用于可复制性和进一步研究.
相关概念视频
00:49Ligand Binding and Linkage
Allosteric proteins have more than one ligand binding site; the binding of a ligand to any of these sites influences the binding of ligands to the other sites. When a protein is allosteric, its binding sites are called coupled or linked. In the case of enzymes, the site that binds to the substrate is known as the active site and the other site is known as the regulatory site. When a ligand binds to the regulatory site, this leads to conformational changes in the protein that can influence the...
02:07Improving Translational Accuracy
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
00:49Ligand Binding and Linkage
Allosteric proteins have more than one ligand binding site; the binding of a ligand to any of these sites influences the binding of ligands to the other sites. When a protein is allosteric, its binding sites are called coupled or linked. In the case of enzymes, the site that binds to the substrate is known as the active site and the other site is known as the regulatory site. When a ligand binds to the regulatory site, this leads to conformational changes in the protein that can influence the...
02:07Improving Translational Accuracy
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...