2022 Fiscal Year Final Research Report
Development and Applications of Nonlinear Dimension Reduction with Weak Supervisiors
Project/Area Number |
17K08235
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Multi-year Fund |
Section | 一般 |
Research Field |
Physical pharmacy
|
Research Institution | Osaka University |
Principal Investigator |
TAKAGI Tatsuya 大阪大学, 大学院薬学研究科, 特任教授 (80144517)
|
Project Period (FY) |
2017-04-01 – 2023-03-31
|
Keywords | FMO法 / IFIE / 化学記述子 / 生理活性空間 / 多様体学習 |
Outline of Final Research Achievements |
Inter Fragment Interaction Energy (IFIE) is a useful piece of information provided by the FMO method, which shows which residues and substrates (e.g. pharmaceuticals) are interacting in an important way. Linear regression with the pIC50 of the drug substance gives the importance of the residue as a regression coefficient. However, the introduction of chemical descriptors is necessary to correct the tendency of FMOs to overestimate the entropy term and move away from electrostatic interactions. In this task, the chemical space (chemical descriptors) and the bioactive space (IFIE) were regressed separately by PLS regression and further mapped into 2D space by manifold learning to obtain the 2D coordinates of the ligands and chemical descriptors, and the residues and descriptors were selected to obtain the best evaluation index.
|
Free Research Field |
計量薬学
|
Academic Significance and Societal Importance of the Research Achievements |
この課題の成果は、COVID-19パンデミックなどもありまだ十分に公開できていないが、成果が広く共有されれば、FMO計算の結果がpIC50とより良い相関を示し、FMO計算から直接、新たな医薬品の開発に繋がる情報を得ることができる。今回のパンデミックで判明したように、医薬品の開発は時として十分な時間を与えられない。そのような場合、医薬品開発の時間短縮に占める本法の意義は大きいと考える。 又、学術的には化学空間と生物空間を別々にPLS回帰し、その後多様体学習を用いることにより、より互いからの「雑音」を消去する形での情報抽出が可能になったことがあげられる。
|