2022 Fiscal Year Final Research Report
Development of 3D model construction method using structure database and electron microscope images
Project/Area Number |
18K06101
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Multi-year Fund |
Section | 一般 |
Review Section |
Basic Section 43020:Structural biochemistry-related
|
Research Institution | National Institutes for Quantum Science and Technology |
Principal Investigator |
Matsumoto Atsushi 国立研究開発法人量子科学技術研究開発機構, 量子生命科学研究所, 主幹研究員 (10399420)
|
Project Period (FY) |
2018-04-01 – 2023-03-31
|
Keywords | プログラム開発 / ニューラルネットワーク / 電子顕微鏡 / 生体分子 / 立体構造 |
Outline of Final Research Achievements |
In this study, a computer method to select or find a matching or similar three-dimensional structure from a database of biomolecules when given an electron microscope image of a biomolecule was developed. The method used a combination of the microscopy image creation technique developed by the lead author and neural networks. For this purpose, a dataset of electron microscopy images was created for machine learning from approximately 20,000 large biomolecules selected from the current database consisting of approximately 200,000 biomolecules. The structures were classified by size and machine learning was performed in each class. In the largest structure class, the accuracy rate was about 70% (with the correct answer included in the top three answers about 90% of the time).
|
Free Research Field |
生物物理
|
Academic Significance and Societal Importance of the Research Achievements |
本研究で開発した計算機手法を用いることにより、立体構造が分かっていない生体分子であっても、それに類似する構造がデータベースに登録されていれば、その電子顕微鏡画像をもとに、立体構造を速やかに類推することができる。 電子顕微鏡による立体構造解析では、通常クライオ電子顕微鏡が用いられるが、非常に高価で、これを持つ施設も限られている。本手法を用いることで、従来の電子顕微鏡を活用して、立体構造解析を行うことができる。
|