药物发现中的变压器:微调ChemBERTa以高精度预测可溶性,毒性和结合亲和力
Sandhya Alagarsamy1, Chin-Shiuh Shieh2, Mong-Fong Horng3
1Department of Computer Science and Engineering, SRM Institute of Science and Technology, Ramapuram, Chennai 60089, Tamil Nadu, India.
Drug discovery today
|January 14, 2026
概括
一个新的AI模型准确地预测了分子特性,提高了药物发现效率. 这种基于变压器的方法提高了预测准确性,降低了制药和农药化合物的成本和临床试验失败.
科学领域:
- 化学中的人工智能.
- 药物发现和开发 药物发现和开发
- 计算化学计算化学
背景情况:
- 药物发现是昂贵和耗时的,许多临床试验由于不准确的分子性质预测而失败.
- 现有的计算模型经常与复杂的化学结构作斗争,限制了它们的有效性.
研究的目的:
- 开发和验证一种新的AI技术,用于准确的分子性质预测.
- 提高药物发现过程的效率和成功率.
主要方法:
- 微调基于变压器的模型ChemBERTa.
- 实现混合架构,将ChemBERTa与图形神经网络结合起来.
- 进行复杂化学结构的系统错误分析.
主要成果:
- 这种混合型号在与MolBERT等传统网络相比,表现出了更高的性能.
- 统计测试证实了该模型增强的预测准确性.
- 该技术对于药物和非药物化合物,包括农药行业的药物和非药物化合物,都被证明是有效的.
结论:
- 人工智能,特别是微调的ChemBERTa模型,可以显著提高分子性质预测的准确性.
- 这种方法有可能降低药物发现成本并提高临床成功率.
- 该模型在各种化学行业的有效性突显了其广泛的适用性.
相关概念视频
The Equilibrium Binding Constant and Binding Strength
14.9K
The equilibrium binding constant (Kb) quantifies the strength of a protein-ligand interaction. Kb can be calculated as follows when the reaction is at equilibrium:
14.9K
Protein-Drug Binding: Determination Methods
616
Determining protein-drug binding can be achieved through indirect and direct methods, each providing valuable insights into the interaction between proteins and drugs.
Indirect methods involve isolating the bound drug from its free form in biological samples such as blood, serum, or plasma. These techniques aim to measure the percentage of drugs bound to proteins. Equilibrium dialysis is a commonly used method where the free drug concentration at equilibrium is measured by separating the bound...
Indirect methods involve isolating the bound drug from its free form in biological samples such as blood, serum, or plasma. These techniques aim to measure the percentage of drugs bound to proteins. Equilibrium dialysis is a commonly used method where the free drug concentration at equilibrium is measured by separating the bound...
616
Conserved Binding Sites
5.0K
Many proteins’ biological role depends on their interactions with their ligands, small molecules that bind to specific locations on the protein known as ligand-binding sites. Ligand-binding sites are often conserved among homologous proteins as these sites are critical for protein function.
Binding sites are often located in large pockets, and if their location on a protein’s surface is unknown, it can be predicted using various approaches. The energetic method computationally...
Binding sites are often located in large pockets, and if their location on a protein’s surface is unknown, it can be predicted using various approaches. The energetic method computationally...
5.0K


