SynLlama:用大型语言模型生成可合成的分子及其类似物
Kunyang Sun1,2,3,4, Dorian Bagni1,2,3,4, Joseph M Cavanagh1,2,3,4
1†Kenneth S. Pitzer Theory Center and Department of Chemistry, ‡Department of Bioengineering, and §Department of Chemical and Biomolecular Engineering, University of California, Berkeley, California 94720, United States.
ACS central science
|December 4, 2025
概括
使用微调的大型语言模型 (LLM) 的新方法SynLlama,为新型分子生成实用的合成途径. 这种生成人工智能工具通过专注于可合成化合物来增强化学空间探索.
科学领域:
- 人工智能的人工智能
- 计算化学计算化学
- 药物发现 药物发现 药物发现
背景情况:
- 生成型机器学习模型为化学空间探索提供了潜力.
- 一个关键的限制是频繁生成具有具有挑战性的合成路径的分子.
- 这阻碍了在药物发现和开发中的实际应用.
研究的目的:
- 开发一种新的生成模型,用于创建可合成的分子.
- 为化学合成规划微调大型语言模型 (LLM).
- 为了创建SynLlama,它能够生成完整的合成路径.
主要方法:
- 微调Meta的Llama3大型语言模型 (LLM).
- 开发SynLlama以使用可访问的构建模块和反应模板生成合成路径.
- 在前进和自下而上的综合计划中评估绩效.
主要成果:
- SynLlama有效地为新型分子生成完整的合成途径.
- 该模型探索了一个可以合成的大型化学空间,数据需求减少.
- SynLlama 展示了对未见的,可购买的构建块的概括.
- 与最先进的方法相比,在合成规划方面表现强.
结论:
- SynLlama提供了一个强大的工具,用于探索可合成的化学空间.
- 该模型提供了实用的合成路线,克服了以前生成方法的局限性.
- SynLlama对于制药应用非常有价值,包括模拟合成和命中扩张.
相关概念视频
Synthetic Biology
5.5K
Synthetic biology is an interdisciplinary science that involves using principles from disciplines such as engineering, molecular biology, cell biology, and systems biology. It involves remodeling existing organisms from nature or constructing completely new synthetic organisms for applications such as protein or enzyme production, bioremediation, value-added macromolecule production, and the addition of desirable traits to crops, to name a few.
Golden rice
Golden rice is a genetically modified...
Golden rice
Golden rice is a genetically modified...
5.5K
ATP and Macromolecule Synthesis
6.8K
Biological macromolecules are organic compounds, predominantly composed of carbon atoms. The carbon atoms are covalently bonded with hydrogen, oxygen, nitrogen, and other minor elements. There are four major biological macromolecule classes: carbohydrates, lipids, proteins, and nucleic acids.
Most macromolecules are composed of single subunits, or building blocks, called monomers. The monomers combine with each other using covalent bonds to form larger molecules known as polymers.
Conversion of...
Most macromolecules are composed of single subunits, or building blocks, called monomers. The monomers combine with each other using covalent bonds to form larger molecules known as polymers.
Conversion of...
6.8K
Lagging Strand Synthesis
16.2K
16.2K
Lagging Strand Synthesis
60.8K
During replication, the complementary strands in double-stranded DNA are synthesized at different rates. Replication first begins on the leading strand. Replication starts later, occurs more slowly, and proceeds discontinuously on the lagging strand.
There are several major differences between synthesis of the leading strand and synthesis of the lagging strand. 1) Leading strand synthesis happens in the direction of replication fork opening, whereas lagging strand synthesis happens in the...
There are several major differences between synthesis of the leading strand and synthesis of the lagging strand. 1) Leading strand synthesis happens in the direction of replication fork opening, whereas lagging strand synthesis happens in the...
60.8K
Molecular Models
43.4K
Physical models representing molecular architectures of chemical compounds play essential roles in understanding chemistry. The use of molecular models makes it easier to visualize the structures and shapes of atoms and molecules.
43.4K
Improving Translational Accuracy
3.5K
3.5K


