数据驱动的方法用于识别超参数在多步复合复合
Annie M Westerlund1, Bente Barge1,2, Lewis Mervin3
1Molecular AI, Discovery Sciences, R&D, AstraZeneca, Gothenburg, Sweden.
优化蒙特卡洛树搜索 (MCTS) 超参数可以提高多步逆合成性能. 数据驱动的方法确定了提高路线可解决性和减少搜索时间的设置,优于当前的默认设置.
科学领域:
- 计算化学计算化学
- 化学中的人工智能.
- 复杂合成规划 复杂合成规划
背景情况:
- 多步复合对于药物发现和化学合成至关重要.
- 复合计划算法的效率,如蒙特卡洛树搜索 (MCTS),高度依赖于超参数调整.
- 在计算逆合成中,平衡搜索时间和路线可解决性是计算逆合成的一个关键挑战.
研究的目的:
- 研究MCTS超参数对多步逆合成性能的影响.
- 为了确定最佳的MCTS超参数设置,以提高反合成的速度和精度.
- 将数据驱动的超参数优化策略与默认设置进行比较.
主要方法:
- 通过系统的网格搜索,贝叶斯优化和机器学习方法评估了MCTS超参数 (代,深度,宽度).
- 使用线性综合速度精度得分 (LISAS) 和反向效率得分来评估性能.
- 在专有和公共数据集 (ChEMBL) 上测试了优化的超参数.
主要成果:
- 确定了一组超参数,其性能明显优于默认AiZynthFinder设置.
- 在内部数据集上实现了93%的可解决性,平均搜索时间为151秒.
- 在ChEMBL数据集上,平均搜索时间为114s,达到74%的可偿债率,超过了默认性能.
结论:
- 数据驱动的超参数优化有效地提高了多步逆合成的MCTS性能.
- 识别的最佳设置为回复合成规划提供了速度和可解决性的优越平衡.
- 动态超参数预测显示了化学合成中实时优化的前景.
更多相关视频
05:57Author Spotlight: In Silico Creation and Impact of Carbonylated Amino Acids on Protein Structure and Function
Published on: April 26, 2024
11:25Multi-step Preparation Technique to Recover Multiple Metabolite Compound Classes for In-depth and Informative Metabolomic Analysis
Published on: July 11, 2014
相关概念视频
Multi-Step Reactions
Rate-Determining Steps
In a multistep reaction mechanism, one of the elementary steps progresses significantly slower than the others. This slowest step is called the rate-limiting step (or rate-determining step). A reaction cannot proceed faster than its slowest step, and hence, the rate-determining step limits the overall reaction rate.
The concept of rate-determining step can be understood from the analogy of a 4-lane freeway with a short-stretch of traffic-bottleneck caused due to...
Structure-Activity Relationships and Drug Design
SAR studies the intricate relationship between a drug's chemical structure and biological activity. It focuses on understanding how modifications to a drug's structure can influence...
Chemical Equilibria: Systematic Approach to Equilibrium Calculations
The first step is to identify all the chemical reactions involved, The...
Diels–Alder vs Retro-Diels–Alder Reaction: Thermodynamic Factors
Predicting Reaction Outcomes
