推进化学安全预测:一个集成的GNN框架与DFT增强的循环化合物溶液.
Seul Lee1,2, Jooyeon Lee3,2, Unghwi Yoon4,2,5
1Department of Statistics, Seoul National University, Seoul, 08826, South Korea.
Journal of cheminformatics
|January 29, 2026
概括
这项研究使用图形神经网络 (GNN) 准确预测关键化学安全特性,如燃烧热量 (HoC),蒸汽压力 (VP) 和闪点. 开发的系统为化学安全评估和应急响应提供了统一的实时解决方案.
科学领域:
- 计算化学和化学信息学
- 机器学习用于材料科学科学
- 化学工程和安全与安全.
背景情况:
- 评估快速扩散的化学物质的安全关键的物理化学性质是具有挑战性的.
- 预测燃烧热量 (HoC),蒸汽压力 (VP) 和闪点等属性的现有方法具有局限性.
- 为了同时预测多个安全性质,需要采用统一而强大的方法.
研究的目的:
- 开发一种基于集成图形神经网络 (GNN) 的方法,用于预测HOC,VP和闪点.
- 提高对具有挑战性的化学结构,特别是循环化合物的预测准确度.
- 创建一个实时预测系统,用于化学安全评估和应急响应的实际应用.
主要方法:
- 在综合数据集上使用图形神经网络 (GNN) 开发了一种统一的预测模型.
- 实施了一种混合方法,将DFT计算和循环化合物的随机森林建模结合起来.
- 将预测模型集成到使用Flask的实时系统中,支持SMILES标记和结构绘图.
主要成果:
- 实现了高预测准确度,平均绝对误差为126J/mol的HOC (R2=0.993),0.617日志单位的VP (R2=0.898),和14.42°C的闪点 (R2=0.839).
- 使用专门的混合方法显著改善了使用0.918的R2周期性化合物的HoC预测.
- 实时系统允许输入化学结构,并与实验数据和基准进行比较.
结论:
- 集成的GNN框架为预测多个安全关键性质提供了强大且统一的解决方案.
- 对循环化合物的专门处理提高了机器学习模型在化学安全方面的可靠性.
- 实时预测系统为化学安全评估和应急响应计划提供了实际实用性.
更多相关视频
10:21Author Spotlight: Streamlining Protein Target Prediction and Validation via Molecular Docking and CETSA
Published on: February 23, 2024
3.7K
09:01A High-throughput Assay for the Prediction of Chemical Toxicity by Automated Phenotypic Profiling of Caenorhabditis elegans
Published on: March 14, 2019
7.7K
相关概念视频
Chemical Reactions in Aqueous Solutions
72.3K
Chemical substances interact in many different ways. Certain chemical reactions exhibit common patterns of reactivity. Due to the vast number of chemical reactions, it becomes necessary to classify them based on the observed patterns of interaction.
72.3K
Stereoisomerism of Cyclic Compounds
11.1K
In this lesson, we delve into the role of ring conformation and its stability, which determines the spatial arrangement and, consequently, the molecular symmetry and stereoisomerism of cyclic compounds. 1,2-Dimethylcyclohexane is used as a case study to evaluate the possible number of stereoisomers. Here, given the multiple (n = 2) chiral centers, there are 2n = 4 possible configurations that lack a plane of symmetry, as the ring skeleton exists in a non-planar chair conformation. In addition,...
11.1K
Relation of DFT to z-Transform
822
The Discrete Fourier Transform (DFT) is a crucial tool for analyzing the frequency content of discrete-time signals. It converts a sequence of N samples from the time domain into its corresponding sequence in the frequency domain, where each sample represents a specific frequency component.
To understand how the DFT works, it's helpful to consider the z-transform, which is a method for representing discrete sequences in the complex frequency domain. The z-transform involves summing the...
To understand how the DFT works, it's helpful to consider the z-transform, which is a method for representing discrete sequences in the complex frequency domain. The z-transform involves summing the...
822
Electrolyte and Nonelectrolyte Solutions
71.8K
Substances that undergo either a physical or a chemical change in solution to yield ions that can conduct electricity are called electrolytes. If a substance yields ions in solution, that is, if the compound undergoes 100% dissociation, then the substance is a strong electrolyte. Complete dissociation is indicated by a single forward arrow. For example, water-soluble ionic compounds like sodium chloride dissociate into sodium cations and chloride anions in aqueous solution.
71.8K
Types of Chemical Bonds
94.2K
Chemical bonding theories were pioneered by American chemist Gilbert N. Lewis. He developed a model called the Lewis model to explain the type and formation of different bonds. Chemical bonding is central to chemistry; it explains how atoms or ions bond together to form molecules. It explains why some bonds are strong and others are weak, or why one carbon bonds with two oxygens and not three; why water is H2O and not H4O.
94.2K
Molecules and Compounds
68.6K
Atoms and Molecules
68.6K
