Presentation Information

10:00 AM - 10:15 AM JST(1:00 AM - 1:15 AM UTC)

[10a-C213-5]Data Analysis of Protein-Ligand Interactions via Tensor Decomposition

〇Hideo Doi¹, Seitaro Okutani¹, Mitsumi Onodera¹, Daiki Arai¹, Eiji Yamamoto², Yoshinori Hirano², Kenji Yasuoka², Shigenori Tanaka³, Kaori Fukuzawa⁴, Yuji Mochizuki^1,5 (1.Rikkyo Univ., 2.Keio Univ., 3.Kobe Univ., 4.Osaka Univ., 5.Univ. Tokyo)

Keywords:

FMO method,Tensor decomposition,Interaction energy

With the advancement of supercomputers like "Fugaku," efficient methods for analyzing massive interaction energy data obtained from the MD-FMO approach are highly demanded. In this study, we developed an analytical protocol by constructing a 3D tensor (ligands × residues × time) from the MD-FMO data of 13 ligand complexes and applying tensor decomposition (CP/Tucker decomposition). As a result, major physical features were extracted with a reproducibility of over 90%, enabling the characterization and identification of ligand similarities based on dynamic interaction patterns.

Back to Session information