[15p-WL1_301-1~9] Application of informaics technology in the field of nanobiotechnology

3:25 PM - 3:40 PM JST(6:25 AM - 6:40 AM UTC)

[15p-WL1_301-5] Prediction of protein interaction energies based on the SCOP2 database

〇Hideo Doi¹, Ryohei Yoshine¹, Takaya Daisuke², Kaori Fukuzawa², Yoshiharu Mori⁷, Koichiro Kato³, Shigenori Tanaka⁴, Koji Okuwaki^1,5, Yuji Mochizuki^1,6 (1.Rikkyo Univ., 2.Osaka Univ., 3.Kyushu Univ., 4.Kobe Univ., 5.JSOL Corp., 6.Univ. Tokyo, 7.Kyushu Univ.)

Keywords:

FMO,Machine learning,Interaction energy

Based on the SCOP2 protein structure classification database, we constructed a machine learning model to predict interaction energies between amino acid residues. In recent years, quantum chemical calculation data using the fragment molecular orbital method has been made publicly available for representative structures in SCOP2; however, this method faces the challenge of enormous computational cost. In this study, we attempted to create a surrogate model for predicting interaction energies by combining the publicly available dataset with low-level quantum chemical calculations.