Methods based on heterogeneous mixed effects model for protein dynamics analysis

2022 Fiscal Year Final Research Report

• Project/Area Number: 19K12203
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 62010:Life, health and medical informatics-related
• Research Institution: Tottori University
• Principal Investigator: AMISAKI Takashi 鳥取大学, 医学部, 教授 (20231996)
• Project Period (FY): 2019-04-01 – 2023-03-31
• Keywords: 変量効果モデル / 構造重ね合わせ

Outline of Final Research Achievements

A method based on a random effects model was developed for estimating covariance matrices that represent the inter- and intra-ensemble variations of protein conformations. The method is applicable under situations such that the dataset is a collection of ensembles each of which is a set of conformations represented in the Cartesian coordinates system. A two-stage method based on the least squares method under heteroscedastic variances was developed as well. Efficiencies of these methods, in particular that of the random effects model at a small ensemble size, were confirmed by numerical tests on machine-synthesized datasets. The estimated principal components for 57 X-ray structures of a kinase were consistent with the functional motifs reported previously. In addition, a combination of the methods with structural clustering was useful for identifying remarkable conformations.

Free Research Field

構造インフォマティクス

Academic Significance and Societal Importance of the Research Achievements

集団間変動と集団内変動を分離するアイデアは従来よりあったが、デカルト座標系における構造重ね合わせから共分散行列の推定までもを統一した手法はこれまでみられていない。とくに、変量効果モデル法では、他集団の情報をベイズ的に考慮しているので、集団サイズが小さい場合に有効である。このため、データベース上の既知の結晶構造群データからそのタンパク質の構造変動や構造異質性についての知見を得るというような、構造生物学関連での研究で活用されることを期待している。