CIPHER: curated single-cell perturbation datasets — main

Curated base Perturb-seq objects underlying the CIPHER analyses (Kuznets-Speck et al., Fluctuation structure predicts genome-wide perturbation outcomes). This record holds the lab-generated and re-curated datasets that have no other public home; the already-public datasets CIPHER uses (Frangieh, Norman, Replogle essential/rpe1, Tian, sci-Plex 3, GSE264667) are not re-hosted here and are fetched from their original repositories by resources/download_resources.py. Format. Each .h5ad is an AnnData object readable with cipher.load_dataset(path): var_names are gene symbols, obs['perturbation'] holds the perturbation label with 'control' marking control cells (proper_filtered.h5ad uses 'negative'), and X is raw counts. The two astrocyte CSVs are MERFISH measurements plus their perturbation design table. Regenerable caches are deliberately excluded. The covariance/forward precomputes the supplementary notebooks read are rebuilt from these files with the generators in notebooks/src/; see the repository README. Files filesizecontentsmd5Marson2025_D1_Rest_filtered.h5ad1.53 GB335,298 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int1670a49f15fddf62ab767188df0df112deMarson2025_D1_Stim48hr_filtered.h5ad2.25 GB270,659 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int16fdd0c99edf47db174412ad721fe547ddMarson2025_D1_Stim8hr_filtered.h5ad1.41 GB265,455 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int16c751deade135a04cb070c67dad532496Marson2025_D2_Rest_filtered.h5ad2.29 GB479,555 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int16474833a3f246a748b8076cdcbdcf63e1Marson2025_D2_Stim48hr_filtered.h5ad3.47 GB473,318 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int1624bcb324497d813a73fe85dfc559b8f5Marson2025_D2_Stim8hr_filtered.h5ad2.57 GB467,697 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int1698dab3fe3fc0070f7862050091380c7dMarson2025_D3_Rest_filtered.h5ad2.65 GB423,747 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int16e8c066b72f12caea34d26985f49474eaMarson2025_D3_Stim48hr_filtered.h5ad3.31 GB403,100 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int1650079baba74ca7234d5218ab2dbc30f4Marson2025_D3_Stim8hr_filtered.h5ad2.65 GB353,408 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int16ed690ea5c48cba784c1cdafec8cd9224Marson2025_D4_Rest_filtered.h5ad1.57 GB275,022 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int161eef63cdcecdfeed408f05df701823edMarson2025_D4_Stim48hr_filtered.h5ad2.56 GB293,102 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int168af5688ba53377e7ab2565f87d81bf03Marson2025_D4_Stim8hr_filtered.h5ad1.60 GB241,035 cells x 18,130 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int165a8d0a68b1161ebf38b615362b597978ReplogleWeissman2022_K562_gwps_filtered.h5ad1.36 GB385,848 cells x 8,248 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int161820bd6080c50de4eeef487ee7ee4443XAtlas2025_HCT116_filtered.h5ad2.62 GB377,537 cells x 38,606 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int16c7c45287d87db928197d308f0ac9da3eXAtlas2025_HEK293T_filtered.h5ad4.16 GB504,256 cells x 38,606 genes; 999 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, int16559fa6d0ea6bc062649ac972ec6e3f5bakana_etal_2026_crispra_perturbseq.h5ad1.36 GB20,207 cells x 36,591 genes; 61 perturbations plus control; control is obs['perturbation'] == 'control' (2,660 cells); csr_matrix, float6421ccba8fda1f632a0cfdd0e0cceb5539kaden25_fibroblast_ctrl_10k_min100_greedy_4gb.h5ad3.98 GB185,700 cells x 30,395 genes; 1,757 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, float32c6f3996124ce6760e605ebdb1600f10dkaden25_rpe1_ctrl_10k_min100_greedy_4gb.h5ad3.97 GB118,900 cells x 31,419 genes; 1,089 perturbations plus control; control is obs['perturbation'] == 'control' (10,000 cells); csr_matrix, float3285bf6634340c1502906699cfa268e56emerfish_perturbed_cells_astrocytes.csv8.6 MB14,926 rows x 277 columns; columns: PPP5C, LAPTM4A, CA12 ...5525ffd9614e03cc323fe5b19436b4d4perturbation_design_astrocytes.csv4.0 MB129 rows x 14927 columns; columns: , Cell_1, Cell_2 ...4e1ed53683fd99986f4ed6e0c25990a5proper_filtered.h5ad2.23 GB52,030 cells x 18,099 genes; 192 perturbations plus control; control is obs['perturbation'] == 'negative' (2,280 cells); csr_matrix, float327791a37cb0a129d8011fc490638918d9schemidt_etal_2022_crispra_perturbseq.h5ad364.8 MB57,835 cells x 24,709 genes; 69 perturbations plus control; control is obs['perturbation'] == 'control' (3,724 cells); csr_matrix, float6463ed2ec36456d5ce429a228252ee89eb Paths shown relative to CIPHER_DATA_DIR in resources/zenodo_manifest.csv.