scROMA: pathway-activity simulation data replicates

Supporting data for the scROMA manuscript (Zhubanchaliyev et al.). scROMA is anSVD-based method for pathway-activity inference in single-cell transcriptomics thatprovides per-cell activity scores, per-gene contributions, permutation-basedsignificance, and a batch-aware extension that corrects batch effects within thegene-set subspace rather than across the full transcriptome. This record archives the data artifacts underlying the figures: the pre-generated ground-truth simulation replicates. Contents scROMA_01_simulation_replicates.tar.gz — 20 AnnData (.h5ad) replicates: 10 single-batch (2,000 cells) and 10 multi-batch (3 batches x 1,000 cells), each with 1,000 genes and 50 pathways (20 active, 30 inactive decoys), 5 cell types. Each object carries the observed sparse count matrix (X), cell-type and batch labels (obs), continuous and binary ground-truth pathway-activity matrices (obsm), pathway-gene membership (uns), and the generative parameters as metadata. pathways.gmt and manifest.json accompany each regime. These are the exact inputs to Figure 3 and Supplementary Note 2. scROMA_01_generative_parameters.json — verbatim copy of DEFAULT_SINGLE_KWARGS and DEFAULT_MULTI_KWARGS together with the replicate seed list (42-51). CitationCite this record together with the manuscript and the software record.