single-cell-preprocessing-with-omicverse

To automate single-cell preprocessing, you can run an end-to-end workflow that handles QC, normalization, HVG detection, PCA, neighborhood graph construction, and UMAP/TSNE embeddings in a single step. This process uses OmicVerse and Scanpy to generate notebook-ready outputs for downstream analysis.

For single-cell RNA-seq preprocessing on a GPU, the workflow supports CPU, CPU-GPU mixed, and GPU RAPIDS stacks. It provides environment guidance and data-transfer tips to scale your preprocessing, including QC and dimensionality reduction, efficiently on compatible hardware.

Leiden clustering is integrated into the single-cell preprocessing workflow as an optional final step. After computing the neighborhood graph and embeddings, the workflow can perform Leiden clustering and save the results alongside raw and normalized data for downstream visualization.

Yes, the preprocessing workflow uses Scanpy alongside OmicVerse to perform HVG detection and PCA. It combines these tools to execute defensive checks, data recovery, and standard preprocessing steps, ensuring reproducible single-cell analysis.

Yes, the workflow saves raw and normalized single-cell data, embeddings, and clustering results. These notebook-ready outputs are automatically stored during the preprocessing pipeline, allowing you to directly proceed to downstream visualization and analysis.