正文：
在药物研发领域，逆合成规划扮演着至关重要的角色，它对于理解药物分子的合成路径以及优化药物设计流程至关重要。近日，上海交通大学人工智能研究院的研究团队提出了一种新的单步逆合成预测技术，该技术通过集成无监督的SMILES序列对齐技术，显著提升了化学反应预测的准确性和效率。
研究人员在《Journal of Cheminformatics》上发表了一篇名为《Ualign: pushing the limit of template-free retrosynthesis prediction with unsupervised SMILES alignment》的论文。他们介绍了一套单步逆合成预测流程，该流程能够高效地从已知药物分子逆向推导出其可能的合成路径。
该技术的关键在于无监督的SMILES序列对齐技术。传统的逆合成预测方法往往忽视了反应物和产物之间存在的相同子结构，这限制了模型的预测效率和准确性。而上海交通大学团队提出的SMILES对齐技术，能够利用这些相同的子结构，提高预测的准确性。
在实验中，研究人员证明了该模型在预测逆合成路径方面的有效性，并表明该模型有潜力成为药物发现的有价值的工具。通过这种方法，研究人员不仅能够提高药物分子的逆合成预测精度，还能够加快药物研发的速度，降低研发成本。
此外，该技术还利用了图神经网络来捕捉分子内部的拓扑结构信息，这使得模型能够提供更为强大的分子表征能力。通过对原子和化学键的精确排列，模型能够更有效地识别和复用反应物和产物之间的相同子结构，从而简化了预测过程，提高了预测的准确率。
这项研究不仅对药物研发领域具有重要意义，也为其他需要高效化学反应预测的领域提供了新的解决方案。随着研究的深入，该技术有望在未来为化学合成和材料科学等领域带来更多的创新和突破。
英语如下：
News Title: “Shanghai Jiao Tong University Breakthrough: New Efficient Method for Anticipated Retrosynthesis Prediction”
Keywords: Retrosynthesis, SMILES Alignment, Efficient Prediction
News Content:
Title: Shanghai Jiao Tong University Team Proposes SMILES Alignment Technology to Enhance Efficiency of Chemical Retrosynthesis Prediction
In the field of drug discovery, retrosynthetic planning plays a pivotal role, essential for understanding the synthesis pathways of drug molecules and optimizing the drug design process. Recently, a research team from the Institute of Artificial Intelligence at Shanghai Jiao Tong University proposed a new single-step retrosynthesis prediction technique that significantly improves the accuracy and efficiency of chemical reaction prediction by integrating unsupervised SMILES sequence alignment technology.
The researchers published a paper titled “Ualign: pushing the limit of template-free retrosynthesis prediction with unsupervised SMILES alignment” in the Journal of Cheminformatics. They introduced a single-step retrosynthesis prediction workflow that can efficiently deduce the possible synthetic pathways of known drug molecules.
The key to this technology lies in the unsupervised SMILES sequence alignment technique. Traditional retrosynthesis prediction methods often overlook the common substructures between reactants and products, which limits the efficiency and accuracy of the models. The SMILES alignment technology proposed by the team from Shanghai Jiao Tong University can leverage these common substructures to enhance prediction accuracy.
In experiments, the researchers demonstrated the effectiveness of the model in predicting retrosynthesis pathways and indicated that it has the potential to become a valuable tool in drug discovery. Through this method, researchers can not only improve the accuracy of retrosynthesis prediction for drug molecules but also accelerate drug development and reduce research and development costs.
Additionally, the technology utilizes graph neural networks to capture topological structure information within molecules, enhancing the model’s molecular representation capabilities. By precisely arranging atoms and chemical bonds, the model can more effectively identify and reuse common substructures between reactants and products, simplifying the prediction process and increasing the accuracy of predictions.
This research is not only significant for the drug discovery field but also provides new solutions for other areas that require efficient chemical reaction prediction. As the research progresses, this technology has the potential to bring more innovation and breakthroughs to the fields of chemical synthesis and materials science in the future.
【来源】https://www.jiqizhixin.com/articles/2024-07-30-8
Views: 1