Mizuno Group Logo
MIZUNO GROUP

観測できない構造をどう認識・理解するか?

私たちは, 生命科学における複雑なデータ──オミクス, 画像, 構造, 文献など──の背後に潜むパターンや構造を数理的に捉えることで, 生命現象の理解のあり方を拡張することを目指しています。

医薬品を軸としつつ, 深層学習や統計的手法を用いて, 恣意性を抑えた表現の獲得と可視化を行い, 観測値に埋もれた本質的な構造を引き出す研究に取り組んでいます。

Pattern Recognition of Life Science Data
Overview figure of GLDADec, a marker-gene guided LDA model for bulk gene expression deconvolution

GLDADec: marker-gene guided LDA modeling for bulk gene expression deconvolution

Iori Azuma, Tadahaya Mizuno, Hiroyuki Kusuhara
Briefings in Bioinformatics (2024)

Inferring cell type proportions from bulk transcriptome data is crucial in immunology and oncology. Here, we introduce guided LDA deconvolution (GLDADec), a bulk deconvolution method that guides topics using cell type-specific marker gene names to estimate topic distributions for each sample.

[Paper]
Figure showing Transformer models' difficulty recognizing molecular chirality from string representations

Difficulty in chirality recognition for Transformer architectures learning chemical structures from string representations

Yasuhiro Yoshikai, Tadahaya Mizuno, Shumpei Nemoto & Hiroyuki Kusuhara
Nature Communications (2024)

We found that the Transformer requires particularly long training to learn chirality and sometimes stagnates with low performance due to misunderstanding of enantiomers.

[Paper]
Mizuno Group Members

水野班のメンバーは、教員、学生、技術専門職員で構成されています。