Artificial intelligence for template-free protein structure prediction: a comprehensive review
参考中译:人工智能无模板蛋白质结构预测综述


          

刊名：Artificial Intelligence Review: An International Science and Engineering Journal
作者：Mufassirin M. M. Mohamed（Griffith University）
Newton M. A. Hakim（Griffith University）
Sattar Abdul（Griffith University）
刊号：738LB014/IP
ISSN：0269-2821
出版年：2023
年卷期：2023, vol.56, no.8
页码：7665-7732
总页数：68
分类号：TP18
关键词：Bioinformatics；Protein structure prediction；Machine learning；Deep learning；Search-based optimisation
参考中译：生物信息学；蛋白质结构预测；机器学习；深度学习；基于搜索的优化
语种：eng
文摘：Protein structure prediction (PSP) is a grand challenge in bioinformatics, drug discovery, and related fields. PSP is computationally challenging because of an astronomically large conformational space to be searched and an unknown very complex energy function to be minimised. To obtain a given protein’s structure, template-based PSP approaches adopt a similar protein’s known structure, while template-free PSP approaches work when no similar protein’s structure is known. Currently, proteins with known structures are greatly outnumbered by proteins with unknown structures. Template-free PSP has obtained significant progress recently via machine learning and search-based optimisation approaches. However, very accurate structures for complex proteins are yet to be achieved at a level suitable for effective drug design. Moreover, ab initio prediction of a protein’s structure only from its amino acid sequence remains unsolved. Furthermore, the number of protein sequences with unknown structures is growing rapidly. Hence, to make further progress in PSP, more sophisticated and advanced artificial intelligence (AI) approaches are needed. However, getting involved in PSP research is difficult for AI researchers because of the lack of a comprehensive understanding of the whole problem, along with the background and the literature of all related sub-problems. Unfortunately, existing PSP review papers cover PSP research at a very high level and only some parts of PSP and only from a particular singular viewpoint. Using a systematic approach, this review paper provides a comprehensive survey of the state-of-the-art template-free PSP research to fill this knowledge gap. Moreover, covering required PSP preliminaries and computational formulations, this paper presents PSP research from AI perspectives, discusses the challenges, provides our commentaries, and outlines future research directions.
参考中译：蛋白质结构预测(PSP)是生物信息学、药物开发及相关领域的一大挑战。PSP在计算上具有挑战性，因为要搜索的构象空间非常大，并且需要最小化未知的非常复杂的能量函数。为了获得给定蛋白质的结构，基于模板的PSP方法采用相似蛋白质的已知结构，而无模板PSP方法在没有相似蛋白质结构的情况下工作。目前，结构已知的蛋白质远远多于结构未知的蛋白质。通过机器学习和基于搜索的优化方法，无模板PSP最近取得了重大进展。然而，复杂蛋白质的非常精确的结构还没有达到适合有效药物设计的水平。此外，仅根据蛋白质的氨基酸序列从头算预测蛋白质的结构仍然没有解决。此外，具有未知结构的蛋白质序列的数量正在迅速增长。因此，为了在PSP方面取得进一步的进展，需要更复杂和更先进的人工智能(AI)方法。然而，对于人工智能研究人员来说，参与PSP研究是困难的，因为缺乏对整个问题的全面了解，以及所有相关子问题的背景和文献。不幸的是，现有的PSP综述论文涵盖了PSP研究的非常高的水平，并且只涉及PSP的一些部分，并且仅从特定的单一观点。本文采用系统的方法，对无模板PSP研究的最新进展进行了全面综述，以填补这一知识空白。此外，本文从人工智能的角度介绍了PSP的研究，讨论了挑战，提供了我们的评论，并概述了未来的研究方向。



国家科技图书文献中心  全球文献资源网  京ICP备05055788号-26  机械工业信息研究院 2018-2024