{"id":"arxiv:2610.31454","kind":"preprint","title":"Benchmarking Recent Single-Cell Synthetic Data Generation Methods on the OneK1K Dataset","authors":["Buse Giledereli","Basar Temiz","Amirreza Sattarzadeh Khanehbargh"],"author_count":3,"year":2026,"published":"2026-10-22","venue":null,"cited_by":0,"doi":null,"arxiv_id":"2610.31454","nct_id":null,"source_url":"https://arxiv.org/abs/2610.31454","archive_url":"https://space-kitty.com/papers/arxiv:2610.31454","updated":"2026-10-22","pmid":null,"pmcid":null,"openalex_id":"W7223357350","pdf_url":"https://arxiv.org/pdf/2610.31454","abstract":"Synthetic single-cell RNA-seq data generation offers a promising route for privacy-aware biomedical data sharing, but donor-linked datasets require careful balancing of biological utility and privacy risk. In this work, we benchmark synthetic generation methods for the CAMDA 2026 ELSA Health Privacy Challenge using the donor-split OneK1K dataset, comparing scGFT, scLDM, and Cell2Sentence across statistical fidelity, downstream biological preservation, and membership inference vulnerability. Because full-transcriptome generation is computationally demanding, we focus generative modeling on highly variable genes and evaluate alternative strategies for reconstructing non-HVG expression. Our results show that non-HVG handling is a central determinant of privacy: directly copying non-HVG values leads to severe leakage, whereas Poisson-based reconstruction substantially reduces membership inference risk. We further find that scGFT and scLDM better preserve biological structure than an independent Poisson baseline, while maintaining near-random privacy profiles under appropriate configurations. These findings highlight the importance of evaluating synthetic single-cell data beyond marginal similarity alone, emphasizing robustness, interpretability, and privacy-conscious design choices throughout preprocessing, generation, and post-processing.","abstract_license":"CC0-1.0 (arXiv descriptive metadata)","license":null,"is_open_access":false,"retracted":false,"peer_reviewed":false,"categories":["q-bio.GN"],"details":null,"provenance":"arxiv","attribution":"Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.","topics":[{"slug":"single-cell-spatial","name":"Single-Cell & Spatial Omics","domain_slug":"biotech"}],"cited_in_dossiers":[]}