FedscGen: privacy-aware federated batch effect correction of single-cell RNA sequencing data -- Preprocessed datasets

This dataset accompanies the publication "FedscGen: Privacy-Aware Federated Batch Effect Correction of Single-Cell RNA Sequencing Data" and includes eight single-cell RNA sequencing (scRNA-seq) datasets used to benchmark the FedscGen and scGen methods. The datasets are provided in .h5ad format and include comprehensive metadata necessary for replication and further analysis.

Datasets

We analyze various datasets to compare FedscGen against scGen (centralized) in terms of batch correction. For simplicity, we refer to the dataset by abbreviations:

1. Cell Line (CL):

   • Derived from the 293t_jurkat experiment with three batches: Zheng et al., 2017.

2. Human Dendritic Cells (HDC):

   • scRNA-seq data of human dendritic cells across two batches: Villani et al., 2017.

3. Human Pancreas (HP):

   • Consolidated data from five sources with 14,767 cells each: Baron et al., 2016; Muraro et al., 2016; Segerstolpe et al., 2016; Wang et al., 2016; Xin et al., 2016.

4. Mouse Brain (MB):

   • Merged datasets with 691,600 and 141,606 cells: Saunders et al., 2018; Rosenberg et al., 2018.

5. Mouse Cell Atlas (MCA):

   • Data focusing on 11 cell types from various organs: Han et al., 2018; The Tabula Muris Consortium, 2018.

6. Mouse Hematopoietic Stem and Progenitor Cells (MHSPC):

   • Data from SMART-seq2 and MARS-seq protocols: Nestorowa et al., 2016; Paul et al., 2015.

7. Mouse Retina (MR):

   • Data from two unassociated laboratories with 26,830 and 44,808 cells: Macosko et al., 2015; Shekhar et al., 2016.

8. PBMC (human Peripheral Blood Mononuclear Cell):

   • scRNA-seq data with two batches: Zheng et al., 2017.

Usage Notes: Each dataset is provided in .h5ad format, compatible with common single-cell analysis tools such as Scanpy. Detailed metadata is included within each file.

Keywords: Single-cell RNA sequencing, scRNA-seq, Batch effect correction, Privacy-aware, Federated learning, scGen, FedscGen, Clinical multi-center studies, Genomics, Bioinformatics

Contact: For questions or further information, please contact Mohammad Bakhtiari at mohammad.bakhtiari@uni-hamburg.de.

License: Creative Commons Attribution 4.0 International (CC BY 4.0)