scROMA: derived pathway-activity data for the cystic fibrosis, breast and lung analyses

Derived intermediates supporting scROMA: batch-aware pathway-activity inference and a ground-truth simulation framework for single-cell transcriptomics (Zhubanchaliyev et al.). This record holds the four artifacts too large for GitHub's 100 MB per-file limit; everything else needed to rebuild a figure is in the analysis repository. scROMA_cf_scanvi_integrated.h5ad - scANVI-integrated cystic fibrosis airway atlas (40,709 cells x 28,024 genes), unlocking Fig 4 and Supplementary Fig S5. It is deposited rather than regenerated because it cannot be regenerated: scANVI is stochastic and the producer script sets no random seed. scROMA_lung_maynard2020_3k_cnv.h5ad - Maynard et al. 2020 lung adenocarcinoma, 3k-cell subset with inferCNV output (Fig 7, Supplementary Figs S8/S9). scROMA_lung_scroma_tumor_results.tar.gz and scROMA_breast_results_batch_aware_scroma.tar.gz - per-pathway scROMA outputs; both extract at the repository root. Retrieve everything into the correct paths with python scripts/fetch_zenodo_data.py from the analysis repository. anndata >= 0.12 is required. See README.md in this record for the full schema contract, including the cell-order guarantee binding these files to the vendored ROMA outputs.