Bakhtiari Mohammad, Bonn Stefan, Theis Fabian, Zolotareva Olga, Baumbach Jan
Institute for Computational Systems Biology, University of Hamburg, Hamburg, Germany.
Institute of Medical Systems Bioinformatics, Center for Biomedical AI (bAIome), Center for Molecular Neurobiology (ZMNH), University Medical Center Hamburg-Eppendorf, Hamburg, 20246, Germany.
Genome Biol. 2025 Jul 22;26(1):216. doi: 10.1186/s13059-025-03684-6.
Single-cell RNA-seq data from clinical samples often suffer from batch effects, but data sharing is limited due to genomic privacy concerns. We present FedscGen, a privacy-preserving communication-efficient federated method built upon the scGen model, enhanced with secure multiparty computation. FedscGen supports federated training and batch effect correction workflows, including the integration of new studies. We benchmark FedscGen across diverse datasets, showing competitive performance-matching scGen on key metrics like NMI, GC, ILF1, ASW_C, kBET, and EBM on the Human Pancreas dataset. Published as a FeatureCloud app, FedscGen enables secure, real-world collaboration for scRNA-seq batch effect correction.
来自临床样本的单细胞RNA测序数据常常受到批次效应的影响，但由于基因组隐私问题，数据共享受到限制。我们提出了FedscGen，这是一种基于scGen模型构建的、具有隐私保护且通信高效的联邦方法，并通过安全多方计算进行了增强。FedscGen支持联邦训练和批次效应校正工作流程，包括整合新的研究。我们在不同数据集上对FedscGen进行了基准测试，在人类胰腺数据集上，它在诸如NMI、GC、ILF1、ASW_C、kBET和EBM等关键指标上展现出与scGen相媲美的竞争性能。作为一个FeatureCloud应用发布，FedscGen实现了用于scRNA-seq批次效应校正的安全、真实世界的协作。
