Protocol for automated optimization of feature selection and clustering parameters in single-cell RNA-seq using scAutoTune.

Single-cell RNA sequencing (RNA-seq) analysis requires dataset-specific tuning of feature selection and clustering parameters. Here, we present a protocol for automated optimization of feature selection and clustering parameters in single-cell RNA-seq using scAutoTune. We describe steps for installing scAutoTune, preparing Seurat objects from public single-cell RNA-seq datasets, and performing grid-based parameter sweeping with optional Harmony batch correction. We then detail procedures for computing silhouette-based performance metrics, visualizing generalized additive model (GAM)-smoothed landscapes, and applying optimal settings to generate uniform manifold approximation and projection (UMAP) embeddings and interpretable clusters. For complete details on the use and execution of this protocol, please refer to Sun et al.1.