--- library_name: scvi-tools license: cc-by-4.0 tags: - biology - genomics - single-cell - model_cls_name:SCVI - scvi_version:1.4.2 - anndata_version:0.12.7 - modality:rna - tissue:various - annotated:True --- ScVI is a variational inference model for single-cell RNA-seq data that can learn an underlying latent space, integrate technical batches and impute dropouts. The learned low-dimensional latent representation of the data can be used for visualization and clustering. scVI takes as input a scRNA-seq gene expression matrix with cells and genes. We provide an extensive [user guide](https://docs.scvi-tools.org/en/stable/user_guide/models/scvi.html). - See our original manuscript for further details of the model: [scVI manuscript](https://www.nature.com/articles/s41592-018-0229-2). - See our manuscript on [scvi-hub](https://www.biorxiv.org/content/10.1101/2024.03.01.582887v2) how to leverage pre-trained models. This model can be used for fine tuning on new data using our Arches framework: [Arches tutorial](https://docs.scvi-tools.org/en/stable/tutorials/notebooks/scrna/scarches_scvi_tools.html). # Model Description Tabula Sapiens is a benchmark, first-draft human cell atlas of nearly 500,000 cells from 24 organs of 15 normal human subjects. # Metrics We provide here key performance metrics for the uploaded model, if provided by the data uploader.
Coefficient of variation The cell-wise coefficient of variation summarizes how well variation between different cells is preserved by the generated model expression. Below a squared Pearson correlation coefficient of 0.4 , we would recommend not to use generated data for downstream analysis, while the generated latent space might still be useful for analysis. **Cell-wise Coefficient of Variation**: | Metric | Training Value | Validation Value | |-------------------------|----------------|------------------| | Mean Absolute Error | 1.52 | 1.54 | | Pearson Correlation | 0.82 | 0.81 | | Spearman Correlation | 0.81 | 0.80 | | R² (R-Squared) | 0.44 | 0.40 | The gene-wise coefficient of variation summarizes how well variation between different genes is preserved by the generated model expression. This value is usually quite high. **Gene-wise Coefficient of Variation**: | Metric | Training Value | |-------------------------|----------------| | Mean Absolute Error | 32.40 | | Pearson Correlation | 0.66 | | Spearman Correlation | 0.75 | | R² (R-Squared) | 0.09 |
Differential expression metric The differential expression metric provides a summary of the differential expression analysis between cell types or input clusters. We provide here the F1-score, Pearson Correlation Coefficient of Log-Foldchanges, Spearman Correlation Coefficient, and Area Under the Precision Recall Curve (AUPRC) for the differential expression analysis using Wilcoxon Rank Sum test for each cell-type. **Differential expression**: | Index | gene_f1 | lfc_mae | lfc_pearson | lfc_spearman | roc_auc | pr_auc | n_cells | | --- | --- | --- | --- | --- | --- | --- | --- | | B cell | 0.94 | 0.20 | 0.97 | 0.99 | 0.04 | 0.02 | 54746.00 | | CD4-positive, alpha-beta T cell | 0.95 | 0.27 | 0.77 | 0.97 | 0.31 | 0.11 | 39280.00 | | CD8-positive, alpha-beta T cell | 0.92 | 0.39 | 0.72 | 0.94 | 0.37 | 0.30 | 14682.00 | | plasma cell | 0.84 | 0.32 | 0.91 | 0.98 | 0.22 | 0.06 | 5944.00 | | naive thymus-derived CD4-positive, alpha-beta T cell | 0.89 | 0.78 | 0.59 | 0.89 | 0.22 | 0.22 | 4154.00 | | macrophage | 0.79 | 0.74 | 0.78 | 0.95 | 0.97 | 0.64 | 1794.00 | | natural killer cell | 0.91 | 1.51 | 0.52 | 0.82 | 0.43 | 0.38 | 1579.00 | | T cell | 0.86 | 1.85 | 0.57 | 0.80 | 0.11 | 0.08 | 1281.00 | | mature NK T cell | 0.89 | 1.42 | 0.59 | 0.80 | 0.51 | 0.44 | 1024.00 | | regulatory T cell | 0.85 | 1.74 | 0.58 | 0.79 | 0.28 | 0.19 | 968.00 | | monocyte | 0.75 | 1.04 | 0.71 | 0.93 | 0.43 | 0.32 | 766.00 | | innate lymphoid cell | 0.36 | 2.00 | 0.66 | 0.82 | 0.05 | 0.02 | 745.00 | | classical monocyte | 0.85 | 1.99 | 0.63 | 0.86 | 0.60 | 0.46 | 552.00 | | neutrophil | 0.87 | 4.67 | 0.56 | 0.65 | 0.27 | 0.02 | 303.00 | | intermediate monocyte | 0.81 | 2.90 | 0.63 | 0.82 | 0.24 | 0.04 | 281.00 | | mast cell | 0.79 | 3.68 | 0.62 | 0.69 | 0.28 | 0.02 | 212.00 | | endothelial cell | 0.65 | 1.77 | 0.67 | 0.86 | 0.13 | 0.03 | 196.00 | | myeloid dendritic cell | 0.77 | 3.34 | 0.62 | 0.78 | 0.22 | 0.02 | 122.00 | | CD4-positive, alpha-beta thymocyte | 0.56 | 5.34 | 0.60 | 0.75 | 0.18 | 0.02 | 122.00 | | stromal cell | 0.60 | 3.51 | 0.62 | 0.78 | 0.21 | 0.02 | 111.00 | | non-classical monocyte | 0.77 | 4.40 | 0.60 | 0.65 | 0.29 | 0.02 | 71.00 | | CD8-positive, alpha-beta thymocyte | 0.52 | 6.83 | 0.54 | 0.65 | 0.27 | 0.02 | 59.00 | | erythrocyte | 0.21 | 7.58 | 0.36 | 0.26 | 0.46 | 0.03 | 42.00 | | plasmacytoid dendritic cell | 0.46 | 6.54 | 0.47 | 0.36 | 0.40 | 0.03 | 16.00 | | hematopoietic precursor cell | 0.38 | 6.40 | 0.49 | 0.40 | 0.34 | 0.02 | 12.00 |
# Model Properties We provide here key parameters used to setup and train the model.
Model Parameters These provide the settings to setup the original model: ```json { "n_hidden": 128, "n_latent": 20, "n_layers": 3, "dropout_rate": 0.05, "dispersion": "gene", "gene_likelihood": "nb", "use_observed_lib_size": true, "latent_distribution": "normal", "use_batch_norm": "none", "use_layer_norm": "both", "encode_covariates": true } ```
Setup Data Arguments Arguments passed to setup_anndata of the original model: ```json { "layer": "counts", "batch_key": "donor_assay", "labels_key": "cell_type", "size_factor_key": null, "categorical_covariate_keys": null, "continuous_covariate_keys": null } ```
Data Registry Registry elements for AnnData manager: | Registry Key | scvi-tools Location | |--------------------------|--------------------------------------| | X | adata.layers['counts'] | | batch | adata.obs['_scvi_batch'] | | labels | adata.obs['_scvi_labels'] | | latent_qzm | adata.obsm['scvi_latent_qzm'] | | latent_qzv | adata.obsm['scvi_latent_qzv'] | | minify_type | adata.uns['_scvi_adata_minify_type'] | | observed_lib_size | adata.obs['observed_lib_size'] | - **Data is Minified**: False
Summary Statistics | Summary Stat Key | Value | |--------------------------|-------| | n_batch | 12 | | n_cells | 129062 | | n_extra_categorical_covs | 0 | | n_extra_continuous_covs | 0 | | n_labels | 25 | | n_latent_qzm | 20 | | n_latent_qzv | 20 | | n_vars | 3000 |
Training **Training data url**: Not provided by uploader If provided by the original uploader, for those interested in understanding or replicating the training process, the code is available at the link below. **Training Code URL**: https://github.com/YosefLab/scvi-hub-models/blob/main/src/scvi_hub_models/TS_train_all_tissues.ipynb
# References The Tabula Sapiens Consortium. The Tabula Sapiens: A multiple-organ, single-cell transcriptomic atlas of humans. Science, May 2022. doi:10.1126/science.abl4896