Mixture modeling of regularization terms with optimization sampling strategies and its application to biological large scale data
Project/Area Number |
26330330
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Multi-year Fund |
Section | 一般 |
Research Field |
Life / Health / Medical informatics
|
Research Institution | Kyushu University |
Principal Investigator |
Maruyama Osamu 九州大学, マス・フォア・インダストリ研究所, 准教授 (20282519)
|
Project Period (FY) |
2014-04-01 – 2017-03-31
|
Project Status |
Completed (Fiscal Year 2016)
|
Budget Amount *help |
¥4,810,000 (Direct Cost: ¥3,700,000、Indirect Cost: ¥1,110,000)
Fiscal Year 2016: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Fiscal Year 2015: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000)
Fiscal Year 2014: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
|
Keywords | 正則化 / マルコフ連鎖モンテカルロ法 / タンパク質複合体 / ガウス分布 / ベイズ推定 / タンパク質間相互作用 / 教師付き学習 / べき法則 / 混合正則化 / モデリング / バイオインフォマティクス / 相互排他 / ガウス分布のベイズ推定 / サンプリング / マルコフ連鎖モンテカルロ / 遺伝子発現ネットワーク / ガウシアン・グラフィカル・モデル |
Outline of Final Research Achievements |
Based on regularization modeling and Markov chain Monte Carlo algorithms, we have developed methods for the protein complex prediction problem and the Bayes estimation of Gaussian distributions. Especially, for the protein complex prediction problem, we have empirically shown the effectiveness of a regularization term based on the information of mutually exclusive protein-protein interactions. In addition, we have developed a supervised learning algorithm for protein complexes with 2 or 3 components. Furthermore, we have designed a regularization term for controlling overlaps between predicted complexes, and showed that the new method with that regularization term outperforms others.
|
Report
(4 results)
Research Products
(13 results)