TwistDAN:用于合成可访问性评估的扭曲域对抗网络.
Qahtan Adnan Aljanabi1, Zhijian Huang1, Ziyu Fan1
1School of Computer Science and Engineering, Central South University, Changsha 410083, Hunan, China.
Journal of chemical information and modeling
|February 26, 2026
概括
TwistDAN通过使用域对抗神经网络来改善合成可访问性预测,以便在各种化学库中更好地泛化. 这减少了药物发现中不必要的合成尝试.
科学领域:
- 计算化学的计算化学
- 药用化学 医学化学
- 机器学习 机器学习
背景情况:
- 合成可访问性 (SA) 预测对于指导药物发现中的分子合成至关重要.
- 现有的SA预测器缺乏跨多种化学领域的概括性,限制了它们在虚拟查中的使用.
- 准确的SA预测对于高效的命中识别和领先优化至关重要.
研究的目的:
- 开发一种新的SA预测模型,TwistDAN (扭曲域对抗网络),具有改进的跨域概括.
- 通过对标记和未标记的分子数据进行监督和对抗训练的结合,利用半监督学习.
- 增强SA预测的实用性,用于大型和多样化的分子库的虚拟选.
主要方法:
- 适应域对抗神经网络 (DANN) 用于使用半监督学习进行SA预测.
- 在64万个被标记的分子 (易合成和难合成) 上使用监督学习,在210万个未被标记的SELFIES变体上进行对抗学习.
- 使用相同的2D分子图表表示,图表注意力网络和梯度逆转层用于域不变学习.
主要成果:
- 在严重的域移动下,TwistDAN表现出强大的跨域泛化,AUROC为0.951.
- 在具有结构相似分子的具有挑战性的歧视任务 (AUROC = 0.938) 中获得了高性能.
- 与领先的方法相比,表现出高精度 (0.980),减少了12个百分点的假阳性预测.
结论:
- TwistDAN显著提高了合成可访问性预测模型的概括能力.
- 该模型的高精度最大限度地减少了浪费的合成努力,加速了药物发现管道.
- TwistDAN为药物化学家提供可解释的基于注意力的可视化,有助于合理的药物设计.
相关概念视频
Assessment of Airway, Skin Color, and Use of Accessory Muscles
1.8K
A thorough assessment of respiratory health is paramount in clinical settings to identify and manage respiratory distress and ensure adequate oxygenation. This article elaborates on the critical aspects of respiratory evaluation, including airway assessment, skin color examination, and the observation of accessory muscle use, which are integral to effectively diagnosing and managing patients with respiratory conditions.
Introduction
The initial evaluation of a patient's respiratory system...
Introduction
The initial evaluation of a patient's respiratory system...
1.8K
Learning Disabilities
660
Learning disabilities are cognitive disorders caused by neurological impairments that affect cognitive functions like language and reading, without indicating overall intellectual or developmental challenges. These disabilities differ from global intellectual or developmental disabilities as they are limited to distinct cognitive functions. Common learning disabilities include dysgraphia, dyslexia, and dyscalculia, each of which impacts unique aspects of learning.
Dyslexia
Dyslexia is a...
Dyslexia
Dyslexia is a...
660
Synthetic Disvision of Polynomials
227
Synthetic division is an efficient algorithmic approach for dividing a polynomial by a linear binomial of the form x - c, where c is a real number. This method is helpful due to its streamlined process, which avoids the more cumbersome steps involved in the traditional long division of polynomials. It simplifies computation and serves as a practical tool for evaluating polynomials and identifying their factors.To perform synthetic division, one begins by listing the coefficients of the...
227
