RetroCaptioner:通过对比标题的可学习图形表示,在端到端的回复合成变压器中超出了注意力
Xiaoyi Liu1,2, Chengwei Ai3, Hongpeng Yang4
1School of Chinese Materia Medica, Beijing University of Chinese Medicine, Beijing, 102488, China.
Bioinformatics (Oxford, England)
|September 28, 2024
概括
RetroCaptioner是一个基于变压器的新型框架,通过有效捕获反应转换来增强化学逆合成. 这种先进的方法提高了药物发现的预测准确性和分子有效性.
科学领域:
- 计算化学计算化学
- 化学中的人工智能.
背景情况:
- 回复合成对于识别新型化合物的前体至关重要.
- 变压器模型越来越多地用于自动化回复合成.
- 现有的方法往往无法捕获反应转换细节,限制了准确性.
研究的目的:
- 介绍RetroCaptioner,一个先进的端到端的基于变压器的化学逆合成框架.
- 提高自动化回复合成预测的准确性和适用性.
- 为了有效地捕获反应转换信息和化学上可信的约束.
主要方法:
- 开发了RetroCaptioner,配备一个对比反应中心标题,用于训练双视图注意力模型.
- 利用对比式学习与分子图表表示来实现单步学习过程.
- 集成的单编码器,双编码器和编码器解码器范式,增强SMILES和分子图之间的原子对应.
主要成果:
- 在USPTO-50k数据集上实现了67.2%的top-1精度和93.4%的top-10精度.
- 获得了99.4%的特殊SMILES有效性得分.
- 在为药物protocylol生成合成路径方面表现出可靠性.
结论:
- RetroCaptioner显著推进了自动化工反合成.
- 该框架捕捉反应细节和约束的能力导致了高准确性和有效性.
- 这种方法有望加速药物发现和开发.
相关概念视频
Transformers
1.1K
A device that transforms voltages from one value to another using induction is called a transformer. A transformer consists of two separate coils, or windings, wrapped around the same soft iron core. However, they are electrically insulated from each other.
The iron core has a substantial relative permeability. Therefore, the magnetic field lines generated due to the current in one winding are almost entirely confined within the core, such that the same magnetic flux permeates each turn of both...
The iron core has a substantial relative permeability. Therefore, the magnetic field lines generated due to the current in one winding are almost entirely confined within the core, such that the same magnetic flux permeates each turn of both...
1.1K
[3,3] Sigmatropic Rearrangement of 1,5-Dienes: Cope Rearrangement
2.6K
The Cope rearrangement is classified as a [3,3] sigmatropic shift in 1,5-dienes, leading to a more stable, isomeric 1,5-diene. The reaction involves a concerted movement of six electrons, four from two π bonds and two from a σ bond, via an energetically favorable chair-like transition state.
2.6K
Conjugate Addition (1,4-Addition) vs Direct Addition (1,2-Addition)
3.1K
α,β-Unsaturated carbonyl compounds with two electrophilic sites, the carbonyl carbon, and the β carbon, are susceptible to nucleophilic attack via two modes: conjugate or 1,4-addition and direct or 1,2-addition.
Conjugate addition results in a thermodynamically stable product. The reaction retains the stronger C=O bond at the expense of the weaker C=C π bond. The process is slow as the β carbon is less electrophilic than the carbonyl carbon.
Direct addition products are...
Conjugate addition results in a thermodynamically stable product. The reaction retains the stronger C=O bond at the expense of the weaker C=C π bond. The process is slow as the β carbon is less electrophilic than the carbonyl carbon.
Direct addition products are...
3.1K
RL Circuit without Source
879
When a DC source is suddenly disconnected from an RL (Resistor-Inductor) circuit, the circuit becomes source-free. Assuming the inductor has an initial current denoted as I0, the initial energy stored in the inductor can be determined.
Applying Kirchhoff's voltage law around the loop of the circuit and substituting the voltages across the inductor and resistor yields a first-order differential equation. A logarithmic equation is obtained by rearranging the terms in this equation,...
Applying Kirchhoff's voltage law around the loop of the circuit and substituting the voltages across the inductor and resistor yields a first-order differential equation. A logarithmic equation is obtained by rearranging the terms in this equation,...
879
Synthesis and Decomposition Reactions
32.5K
Synthesis and decomposition are two types of redox reactions. Synthesis means to make something, whereas decomposition means to break something. The reactions are accompanied by chemical and energy changes.
32.5K
Source Transformation for AC Circuits
550
The process of source transformation in the frequency domain entails the conversion of a voltage source, positioned in series with an impedance, into a current source that is parallel to an impedance, or the other way around. It is essential to maintain the following relationships while transitioning from one source type to another.
550


