Data-driven predictive approach for designing drug molecules with Bayesian statistics and quantum chemistry
| Project/Area Number |
15H02672
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Statistical science
|
|---|
| Research Institution | The Institute of Statistical Mathematics |
|---|
Principal Investigator |
Yoshida Ryo 統計数理研究所, データ科学研究系, 教授 (70401263)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
本郷 研太 北陸先端科学技術大学院大学, 情報社会基盤研究センター, 准教授 (60405040)
|
|---|
| Project Period (FY) |
2015-04-01 – 2019-03-31
|
| Project Status |
Completed (Fiscal Year 2018)
|
| Budget Amount *help |
¥16,640,000 (Direct Cost: ¥12,800,000、Indirect Cost: ¥3,840,000)
Fiscal Year 2018: ¥3,640,000 (Direct Cost: ¥2,800,000、Indirect Cost: ¥840,000)
Fiscal Year 2017: ¥3,640,000 (Direct Cost: ¥2,800,000、Indirect Cost: ¥840,000)
Fiscal Year 2016: ¥4,290,000 (Direct Cost: ¥3,300,000、Indirect Cost: ¥990,000)
Fiscal Year 2015: ¥5,070,000 (Direct Cost: ¥3,900,000、Indirect Cost: ¥1,170,000)
|
|---|
| Keywords | 機械学習 / 量子化学計算 / 分子設計 / マテリアルズインフォマティクス / 転移学習 / ベイズ推論 / 量子化学 / 創薬 / シミュレーション / ベイズ統計 |
|---|
| Outline of Final Research Achievements |
Using Bayesian inference as a driving force technology, we have developed a machine learning algorithm for designing molecules. Given a data set from experiments or computer simulation, we derive machine learning models forwardly predicting physicochemical properties of given materials. Inverting such trained forward models through Bayes’ law, we derive a posterior distribution for the backward prediction, which is conditioned by a desired property requirement. Exploring high-probability regions of the posterior with a sequential Monte Carlo technique, molecules that exhibit the desired properties can computationally be created. We have developed open source libraries of R and Python (iQSPR, XenonPy). With this technology, we designed promising hypothetical polymers targeting to achieve high thermal conductivity, and three were selected for monomer synthesis and polymerization. The synthesized polymers reached 80% higher thermal conductivities than conventional commercial polymers.
|
| Academic Significance and Societal Importance of the Research Achievements |
同技術を用いることで，任意の物性をターゲットに大量かつ高品質の候補物質のライブラリを作製できるようになった．さらに，逆合成経路探索の機械学習アルゴリズムや計算機シミュレーションとの循環システムを構築すれば，候補物質の性能検証，合成経路のプランニングまでの工程を完全に自動化できる．解析技術は汎用的であり，薬剤分子のみならず，一般の低分子化合物，高分子，混合材料等，様々な材料系に適用できる．現在，物質科学とデータ科学の融合を図るマテリアルズインフォマティクス（MI）と呼ばれる学際領域に注目が集まっている．本研究がもたらした技術と科学的発見は，分子系材料のMIの学術創生に一石を投じるものである．
|
Report
(5 results)
Research Products
(33 results)
-
[Journal Article] マテリアルズ・インフォマティクス概説2018
Author(s)
吉田 亮, 山田 寛尚, Chang Liu, Zhongliang Guo, Stephen Wu
-
Journal Title
CICSJ Bulletin
Volume: 36
Issue: 1
Pages: 9
DOI
NAID
ISSN
0913-3747, 1347-2283
Related Report
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
