A Reproducible Pipeline for Preprocessing and Annotation of scRNA-seq Data Using Seurat and Scanpy

Abstract

Single-cell RNA sequencing (scRNA-seq) is now a versatile platform for the dissection of cellular heterogeneity across biological conditions. Standardization of preprocessing and annotation pipelines is still to come. We present here a reproducible and modular workflow that combines the strengths of Seurat (R) and Scanpy (Python) to preprocess, annotate, and prepare scRNA-seq data for downstream analysis. The workflow begins with raw count matrices from greater than one biological replicates or conditions. Utilizing Seurat, we perform initial quality control, low-quality cell removal, and reference-based cell type annotation from a reference scRNA-seq atlas. The annotated data is re-coded to AnnData format for an easy transition to the Scanpy framework. In Scanpy, additional operations such as normalization, feature selection, dimensionality re- duction (PCA, UMAP), and checking for batch effects are performed. The output data structure is conducive to flexible downstream analysis, including differential expression and pathway enrichment. This pipeline ensures interoperability, reproducibility, and transparency and is particu- larly suited for group environments and comparative analysis. All of the preprocessing is thoroughly documented and parameterized to be straightforwardly modifiable for a range of datasets and research questions.