Molecular Drug-likeness Assessment with Graph Nerural Network

カテゴリ：部門大会

論文No：OS4-2-7

グループ名：【C】2025年電気学会電子・情報・システム部門大会

発行日：2025/8/20

タイトル(英語)：Molecular Drug-likeness Assessment with Graph Nerural Network

著者名：ZHANG YUNFEI（富山大学）,ZHANG CHAO（富山大学）

著者名(英語)： YUNFEI ZHANG (University of Toyama),CHAO ZHANG (University of Toyama)

キーワード：Drug-likeness Assessment,Decision boundary,Graph Neural Networks,Computational Drug Discovery

要約(日本語)：Drug-likeness assessment plays a critical role in the screening of potential drug candidates. However, this task remains challenging, as drug-likeness is typically determined through clinical trials, which are both time-consuming and costly. Recently, binary classification models based on deep learning have been proposed. However, drug-likeness prediction performance is sensitive to negative sample construction. Here, we propose a GNN-based classification model that is trained exclusively on known drug molecules. The model distinguishes molecular drug-likeness by learning an embedding space derived from known drug molecules. Our model demonstrates relatively consistent performance across diverse datasets. Moreover, it provides a scoring mechanism to quantify drug-likeness.

本誌掲載ページ：972-975p

原稿種別：英語

PDFファイルサイズ：299Kバイト