kMoL:一个开源的机器和联合学习库,用于药物发现
Romeo Cozac1, Haris Hasic2, Jun Jin Choong2
1Elix, Inc., 8-34 Yonbancho, Chiyoda-ku, Tokyo, 102-0081, Japan. romeo.cozac@elix-inc.com.
Journal of cheminformatics
|February 26, 2025
概括
kMoL是一个新的开源药物发现库,它使用联合学习来实现协作模型开发,同时保持数据隐私. 它为制药研究中的机器学习提供了先进的定制和安全性.
科学领域:
- 计算化学和化学信息学
- 机器学习在药物发现中的作用
- 在AI中的数据隐私和安全性.
背景情况:
- 机器学习,特别是图形卷积网络 (GCNs),对于药物发现任务至关重要,例如定量结构-活性关系 (QSAR) 和ADME.
- 数据隐私问题阻碍了在制药行业改善机器学习模型性能和稳健性的协作努力.
研究的目的:
- 介绍和评估kMoL,一个具有集成联合学习能力的开源机器学习库.
- 解决与数据隐私和安全相关的药物发现方面的挑战.
- 提供有关部署保护隐私的机器学习模型的指导.
主要方法:
- 开发了kMoL,这是一个开源库,具有最先进的GCN架构,贝叶斯优化,可解释性和联合学习.
- 通过本地基准培训和分布式联合学习实验对不同数据集进行kMoL评估.
- 评估了联合学习策略的定制,安全,适应性和性能权衡.
主要成果:
- 在没有额外的编程的情况下,kMoL展示了广泛的定制,高级安全性和适应用户特定模型和数据集的适应性.
- 联合学习实验提供了对性能权衡的见解,指导了维护隐私的模型部署.
- kMoL在药物发现中促进了快速和实用的实验.
结论:
- kMoL提供了一个可访问,安全和开源的平台,用于协作药物发现,通过联合学习增强模型开发.
- 该图书馆使研究人员能够构建强大的机器学习模型,同时保持数据隐私.
- 结果为在制药管道中实施保护隐私的人工智能提供了有价值的见解.
相关概念视频
Drug Discovery: Overview
7.4K
Drug discovery is a multifaceted process involving extensive screening, testing, and optimization of lead compounds to identify potential new drugs for therapeutic use. It combines several approaches, including screening large numbers of natural products, chemical modification of known active molecules, identification of new drug targets, and rational design based on biological mechanisms and drug-receptor structure. These approaches are carried out in both academic research laboratories and...
7.4K
Drug-Receptor Interaction: Agonist
2.3K
Agonists are drugs that interact with specific receptors in the body to produce a biological response. When an agonist binds to a receptor, it activates or enhances the receptor's function, leading to physiological effects. The interaction between agonist drugs and receptors is crucial for their therapeutic action in various medical treatments.
Agonists can bind to receptors in different ways. Some agonists bind directly to the receptor's active site, mimicking the endogenous...
Agonists can bind to receptors in different ways. Some agonists bind directly to the receptor's active site, mimicking the endogenous...
2.3K
Pharmacokinetic Models: Overview
543
Pharmacokinetic models utilize mathematical analysis to achieve a detailed quantitative understanding of a drug's life cycle within the body. They are instrumental in simulating a drug's pharmacokinetic parameters, predicting drug concentrations over time, optimizing dosage regimens, linking concentrations with pharmacologic activity, and estimating potential toxicity.
There are three primary types of models: empirical, compartment, and physiological. Empirical models, with minimal...
There are three primary types of models: empirical, compartment, and physiological. Empirical models, with minimal...
543
Quantitative Aspects of Drug-Receptor Interaction
918
The receptor occupancy theory connects a drug's response to the number of occupied receptors. With higher drug concentrations, more receptors are occupied, leading to increased responses. The formation of drug-receptor complexes involves association and dissociation rates, which reach equilibrium when the forward and backward reactions are equal. The equilibrium association constant (Ka) and its inverse, the equilibrium dissociation constant (Kd), indicate drug affinity. Higher Ka and lower...
918
Drug Distribution: Plasma Protein Binding
4.7K
Drugs predominantly attach to plasma proteins, with only a small percentage remaining unbound. The unbound portion can be calculated as one minus the bound fraction. Acidic drugs form large, inactive complexes by reversibly binding to plasma albumin, which prevents them from diffusing across biological barriers. These drug-protein complexes act as reservoirs for the drugs. As the concentration of unbound drugs decreases, these complexes quickly dissociate to release the free drug, maintaining...
4.7K
Model Approaches for Pharmacokinetic Data: Distributed Parameter Models
54
Pharmacokinetic models are mathematical constructs that represent and predict the time course of drug concentrations in the body, providing meaningful pharmacokinetic parameters. These models are categorized into compartment, physiological, and distributed parameter models.
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
54


