RT Journal Article
SR Electronic
T1 geneBasis: an iterative approach for unsupervised selection of targeted gene panels from scRNA-seq
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 2021.08.10.455720
DO 10.1101/2021.08.10.455720
A1 Alsu Missarova
A1 Jaison Jain
A1 Andrew Butler
A1 Shila Ghazanfar
A1 Tim Stuart
A1 Maigan Brusko
A1 Clive Wasserfall
A1 Harry Nick
A1 Todd Brusko
A1 Mark Atkinson
A1 Rahul Satija
A1 John Marioni
YR 2021
UL http://biorxiv.org/content/early/2021/08/10/2021.08.10.455720.abstract
AB The problem of selecting targeted gene panels that capture maximum variability encoded in scRNA-sequencing data has become of great practical importance. scRNA-seq datasets are increasingly being used to identify gene panels that can be probed using alternative molecular technologies, such as spatial transcriptomics. In this context, the number of genes that can be probed is an important limiting factor, so choosing the best subset of genes is vital. Existing methods for this task are limited by either a reliance on pre-existing cell type labels or by difficulties in identifying markers of rare cell types. We resolve this by introducing an iterative approach, geneBasis, for selecting an optimal gene panel, where each newly added gene captures the maximum distance between the true manifold and the manifold constructed using the currently selected gene panel. We demonstrate, using a variety of metrics and diverse datasets, that our approach outperforms existing strategies, and can not only resolve cell types but also more subtle cell state differences. Our approach is available as an open source, easy-to-use, documented R package (https://github.com/MarioniLab/geneBasisR).Competing Interest StatementThe authors have declared no competing interest.