Luecken multiome human bone marrow 2021 s4d8 cellular structure

Dataset published by Luecken et al. 2021 which contains data from human bone marrow measured through joint profiling of single-nucleus RNA and chromatin accessibility using the 10X Genomics Single Cell Multiome ATAC + Gene Expression Kit. All files stored in this entry are from sample 4 donor 8 and contain updated/additional objects depending on the preprocessing step applied. The respective raw files can be downloaded here: File description s4d8_clustered.h5ad: Clustered and normalized cells by features matrix. This file only includes cells from the sample of site 4, donor 8. Cells that did not pass QC are removed. AnnData object includes raw counts (under .X) and normalized counts (under adata.layers['scran_normalization']). reference_embedding.h5ad: 10-dimensional integrated embedding of the dataset as described above (excluding site 4, donor 8 sample) created by integrating with scVI, using site + donor as the batch variable. gene_info.csv: table listing the genes (in order) used for the scVI integration model, and needed for query-to-reference mapping with scArches model.pt: integration scVI model of the bone marrow reference, needed for query-to-reference mapping with scArches Citation Luecken, M. D. et al. A sandbox for prediction and integration of DNA, RNA, and proteins in single cells. Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (2021). Original data link https://openproblems.bio/neurips_docs/data/dataset/