Original title: scSniper: Single-cell Deep Neural Network-based Identification of Prominent Biomarkers
Authors: Mingyang Li,Yanshuo Chen,Jun Ding
In this article, the authors discuss the importance of discovering disease biomarkers at the single-cell level to advance our understanding of diseases and improve diagnostic accuracy. They explain that current computational methods have limitations, such as relying on prior knowledge, being restricted to unimodal data, and using conventional statistical tests for feature selection. To address these issues, the authors introduce a new approach called scSniper.
ScSniper is a novel deep neural network framework that specifically targets robust single-cell multi-omic biomarker detection. One notable feature of scSniper is its mimetic attention block, which improves alignment across different types of data. Additionally, scSniper utilizes sensitivity analysis based on a deep neural network for feature selection and can uncover complex gene regulatory networks without requiring prior knowledge.
The authors evaluate scSniper using real-world datasets, including COVID-19 CITE-Seq and LUAD scRNA-Seq, and show that it consistently outperforms traditional methods like MAST, Wilcox, and DESeq2 in identifying critical biomarkers. The scSniper tool and related experimental codes are publicly available on GitHub. The authors declare no competing interests in this research.
Original article: https://www.biorxiv.org/content/10.1101/2023.11.22.568389v1