Single-cell dataset for non-genetic mechanisms of fractional resistance to abemaciclib in dedifferentiated liposarcoma

Le résumé fourni par la source

Single-cell datasets for Non-Genetic Mechanisms of Fractional Resistance to Abemaciclib in Dedifferentiated Liposarcoma. DDLPS_cell_line_Lipo246_subsampled_data.h5ad: Subsampled dataset of the dedifferentiated liposarcoma (DDLPS) cell line LPS246 across three treatment conditions (control, 50 nM, and 250 nM abemaciclib), 2,000 cells per condition (6,000 cells total). AnnData object format. 81 standardized single-cell 4i features per cell: nuclear morphology (area, orientation, major / minor axis length) and four measurement types per protein — nuclear mean intensity, perinuclear ring mean intensity, total nuclear protein, and nuclear-to-cytoplasmic ratio — for protein channels imaged across seven sequential rounds (R0–R6), with the DNA channel retained only for the final round. Cell metadata includes well, sample ID, treatment, dose, nuclear centroid coordinates, and a pRB high / low proliferation label. DDLPS_cell_line_Lipo246_subsampled_data_PHATE.h5ad: Same 6,000 LPS246 cells as DDLPS_cell_line_Lipo246_subsampled_data.h5ad, restricted to the 18 nuclear-mean protein features used as input to PHATE manifold learning (R0–R6 nuc_mean channels; R6 DNA only). AnnData object format. PHATE coordinates (knn = 100, t = 7) computed on these 18 features are stored in obsm["X_phate"] for manifold visualization. An interactive 3D PHATE viewer is available at https://laelise.github.io/ddlps-fractional-resistance/ DDLPS_primary_tumor_8675CL_subsampled_data.h5ad: Subsampled dataset of a dedifferentiated liposarcoma (DDLPS) primary tumor sample (8675CL) across three treatment conditions (control, 50 nM, and 250 nM abemaciclib), 2,000 cells per condition (6,000 cells total). AnnData object format. 65 standardized single-cell 4i features per cell: nuclear morphology (area, orientation, major / minor axis length) and three measurement types per protein — nuclear mean intensity, perinuclear ring mean intensity, and total nuclear protein — for protein channels imaged across seven sequential rounds (R0–R6), with the DNA channel retained only for the final round. Cell metadata includes well, sample ID, treatment, dose, and nuclear centroid coordinates. DDLPS_cell_line_Lipo246_Tagto_Abema_cotreatment_triplicate.h5ad: Co-treatment triplicate dataset of the dedifferentiated liposarcoma (DDLPS) cell line LPS246 across a 7×3 dose-response matrix of tagtociclibe (CDK2 inhibitor: 0, 0.25, 0.5, 1, 2.5, 5, and 10 µM) crossed with abemaciclib (CDK4/6 inhibitor: 0, 10, and 50 nM), performed in three independent biological replicates (Rep1, Rep2, Rep3). AnnData object format. Standardized single-cell 4i features per cell follow the same panel structure as Further description: This is a single-cell dataset of dedifferentiated liposarcoma (DDLPS) cells gathered using iterative indirect immunofluorescent imaging (4i). Two DDLPS samples were profiled in parallel: an established DDLPS cell line (LPS246) and a primary tumor (8675CL) resected from a consented patient. The two samples were cultured, treated, and imaged independently. Each sample was treated with three conditions of the CDK4/6 inhibitor abemaciclib (control, 50 nM, and 250 nM) and then imaged using 4i to obtain per-cell proteomic measurements at single-cell resolution. In each 4i experiment, cells are imaged across seven sequential rounds (R0–R6), where each round visualizes a fresh panel of antibodies along with a DNA stain used for image registration. Because the DNA channel re-encodes the same biological signal in every round, we retain only the final-round DNA (R6) and drop the earlier-round DNA features. After segmentation, per-cell intensities were extracted for the nuclear mean, perinuclear ring mean, total nuclear protein, and the nuclear-to-cytoplasmic ratio. Feature values are standardized within each sample. From each sample, we then drew a representative downsample of 2,000 cells per treatment condition (6,000 cells per sample) to give an even, manageable view of each dataset. The treatment and dose columns together encode the experimental condition, sample_ID is the unique identifier per treatment group, well refers to the physical well on the multiwell imaging plate, and centroid-0 / centroid-1 give the nuclear centroid position in image coordinates. The pRb_label column is a per-cell proliferation annotation found by selecting the upper modality of pRb / Rb values: pRB_high cells are still proliferating, while pRB_low cells have arrested. For LPS246, we additionally provide a companion file with a 3D PHATE manifold (knn = 100, t = 7) computed on the 18 nuclear-mean protein features, stored in obsm["X_phate"], for figure reproduction and manifold visualization.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.