Turn bulk RNA-seq cohorts into synthetic single-cell datasets using omicverse's Bulk2Single workflow for cell fraction estimation, beta-VAE generation, and quality control comparisons against reference scRNA-seq.
Use this skill when a user wants to reconstruct single-cell profiles from bulk RNA-seq together with a matched reference scRNA-seq atlas. It follows t_bulk2single.ipynb, which demonstrates how to harmonise PDAC bulk replicates, train the beta-VAE generator, and benchmark the output cells against dentate gyrus scRNA-seq.
omicverse as ov, scanpy as sc, scvelo as scv, anndata, and matplotlib.pyplot as plt, then call ov.plot_set() to match omicverse styling.ov.read(...)/ov.utils.read(...) and harmonise gene identifiers via ov.bulk.Matrix_ID_mapping(<df>, 'genesets/pair_GRCm39.tsv').scv.datasets.dentategyrus()adata.obs['clusters']ov.bulk2single.Bulk2Single(bulk_data=bulk_df, single_data=adata, celltype_key='clusters', bulk_group=['dg_d_1', 'dg_d_2', 'dg_d_3'], top_marker_num=200, ratio_num=1, gpu=0).gpu=-1 forces CPU) and how bulk_group names align with column IDs in the bulk matrix.model.predicted_fraction() to run the integrated TAPE estimator, then plot stacked bar charts per sample to validate proportions.df.to_csv(...)).model.bulk_preprocess_lazy(), model.single_preprocess_lazy(), and model.prepare_input() to produce matched feature spaces.model.train(batch_size=512, learning_rate=1e-4, hidden_size=256, epoch_num=3500, vae_save_dir='...', vae_save_name='dg_vae', generate_save_dir='...', generate_save_name='dg').patience and how to resume by reloading weights with model.load('.../dg_vae.pth').model.plot_loss() to monitor convergence.model.generate() and reduce noise through model.filtered(generate_adata, leiden_size=25)..write_h5ad) for reuse, noting it contains PCA embeddings in obsm['X_pca'].ov.bulk2single.bulk2single_plot_cellprop(...) for both generated and reference data.ov.bulk2single.bulk2single_plot_correlation(single_data, generate_adata, celltype_key='clusters').generate_adata.obsm['X_mde'] = ov.utils.mde(generate_adata.obsm['X_pca']) and visualise via ov.utils.embedding(..., color=['clusters'], palette=ov.utils.pyomic_palette()).top_marker_num or provide a curated marker list.bulk_group names—double-check column IDs in the bulk matrix.gpu to an available CUDA device for speed.t_bulk2single.ipynbomicverse_guide/docs/Tutorials-bulk2single/data/reference.md