README.md

---
library_name: scvi-tools
license: cc-by-4.0
tags:
- biology
- genomics
- single-cell
- model_cls_name:TOTALVI
- scvi_version:1.2.0
- anndata_version:0.11.1
- modality:rna
- modality:protein
- tissue:thymus
- annotated:True
---


TotalVI is a variational inference model for single-cell RNA-seq as well as protein data that can
learn an underlying latent space, integrate technical batches, impute dropouts,
and predict protein expression given gene expression or missing protein data given gene expression
and protein data for a subset of proteins.
The learned low-dimensional latent representation of the data can be used for visualization and
clustering.

TotalVI takes as input a scRNA-seq gene expression and protein expression matrix with cells and
genes.
We provide an extensive [user guide](https://docs.scvi-tools.org/en/1.2.0/user_guide/models/totalvi.html).

- See our original manuscript for further details of the model:
[TotalVI manuscript](https://www.nature.com/articles/s41592-020-01050-x).
- 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/1.0.0/tutorials/notebooks/scarches_scvi_tools.html).


# Model Description

CITE-seq to measure RNA and surface proteins in thymocytes from wild-type and T cell lineage-restricted mice to generate a comprehensive timeline of cell state for each T cell lineage.

# Metrics

We provide here key performance metrics for the uploaded model, if provided by the data uploader.

<details>
<summary><strong>Coefficient of variation</strong></summary>

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**:

Modality: rna

| Metric                  | Training Value | Validation Value |
|-------------------------|----------------|------------------|
| Mean Absolute Error | 2.12  | 2.10           |
| Pearson Correlation | 0.54  | 0.61  |
| Spearman Correlation | 0.62 | 0.61  |
| R² (R-Squared) | -3.43  | -5.98      |

Modality: protein

| Metric                  | Training Value | Validation Value |
|-------------------------|----------------|------------------|
| Mean Absolute Error | 0.23  | 0.23           |
| Pearson Correlation | 0.72  | 0.72  |
| Spearman Correlation | 0.71 | 0.71  |
| R² (R-Squared) | 0.38  | 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**:

Modality: rna

| Metric                  | Training Value |
|-------------------------|----------------|
| Mean Absolute Error | 5.44   |
| Pearson Correlation | 0.85  |
| Spearman Correlation | 0.98 |
| R² (R-Squared) | 0.68  |

Modality: protein

| Metric                  | Training Value |
|-------------------------|----------------|
| Mean Absolute Error | 0.23   |
| Pearson Correlation | 0.93  |
| Spearman Correlation | 0.96 |
| R² (R-Squared) | 0.59  |



</details>

<details>
<summary><strong>Differential expression metric</strong></summary>

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**:

Modality: rna

| Index | gene_f1 | lfc_mae | lfc_pearson | lfc_spearman | roc_auc | pr_auc | n_cells |
| --- | --- | --- | --- | --- | --- | --- | --- |
| DP (Q2) | 0.88 | 2.15 | 0.45 | 0.82 | 0.08 | 0.02 | 10864.00 |
| DP (Sig.) | 0.84 | 2.10 | 0.39 | 0.85 | 0.03 | 0.02 | 9824.00 |
| DP (Q1) | 0.90 | 1.90 | 0.44 | 0.84 | 0.02 | 0.02 | 8556.00 |
| Mature CD4 | 0.93 | 2.20 | 0.55 | 0.88 | 0.05 | 0.02 | 6525.00 |
| Immature CD8 | 0.94 | 2.19 | 0.51 | 0.86 | 0.02 | 0.02 | 5686.00 |
| DP (P) | 0.92 | 1.37 | 0.62 | 0.95 | 0.85 | 0.08 | 5593.00 |
| Immature CD4 | 0.90 | 2.17 | 0.52 | 0.85 | 0.02 | 0.02 | 5164.00 |
| Mature CD8 | 0.87 | 2.38 | 0.51 | 0.87 | 0.34 | 0.07 | 4234.00 |
| DN | 0.80 | 1.59 | 0.57 | 0.93 | 0.68 | 0.13 | 2395.00 |
| GD T | 0.82 | 2.40 | 0.54 | 0.79 | 0.34 | 0.33 | 2279.00 |
| Treg | 0.88 | 2.55 | 0.52 | 0.77 | 0.42 | 0.37 | 1966.00 |
| Neg. sel. (2) | 0.84 | 2.72 | 0.61 | 0.75 | 0.29 | 0.26 | 1560.00 |
| Dying | 0.65 | 3.33 | 0.51 | 0.49 | 0.02 | 0.01 | 1552.00 |
| Neg. sel. (1) | 0.77 | 3.01 | 0.57 | 0.73 | 0.20 | 0.12 | 1206.00 |
| Mature cycling | 0.87 | 2.65 | 0.59 | 0.84 | 0.78 | 0.71 | 992.00 |
| Interferon sig. | 0.85 | 2.93 | 0.63 | 0.64 | 0.26 | 0.15 | 984.00 |
| NKT | 0.65 | 2.21 | 0.52 | 0.77 | 0.58 | 0.40 | 928.00 |
| Myeloid | 0.86 | 1.61 | 0.62 | 0.85 | 0.54 | 0.41 | 908.00 |
| Doublet | 0.51 | 3.48 | 0.49 | 0.39 | 0.30 | 0.03 | 677.00 |
| B | 0.81 | 4.19 | 0.62 | 0.60 | 0.26 | 0.02 | 106.00 |
| Erythrocyte | 0.82 | 4.84 | 0.58 | 0.46 | 0.42 | 0.02 | 43.00 |

Modality: protein

| Index | gene_f1 | lfc_mae | lfc_pearson | lfc_spearman | roc_auc | pr_auc | n_cells |
| --- | --- | --- | --- | --- | --- | --- | --- |
| DP (Q2) | 0.91 | 0.04 | 1.00 | 0.99 | 0.11 | 0.15 | 10864.00 |
| DP (Sig.) | 1.00 | 0.03 | 0.99 | 0.97 | 0.27 | 0.32 | 9824.00 |
| DP (Q1) | 0.91 | 0.04 | 1.00 | 1.00 | 0.11 | 0.15 | 8556.00 |
| Mature CD4 | 1.00 | 0.06 | 1.00 | 1.00 | 0.11 | 0.15 | 6525.00 |
| Immature CD8 | 0.82 | 0.05 | 0.99 | 0.99 | 0.36 | 0.38 | 5686.00 |
| DP (P) | 1.00 | 0.06 | 0.99 | 0.97 | 0.25 | 0.18 | 5593.00 |
| Immature CD4 | 1.00 | 0.04 | 1.00 | 0.98 | 0.27 | 0.27 | 5164.00 |
| Mature CD8 | 1.00 | 0.07 | 0.99 | 0.98 | 0.09 | 0.15 | 4234.00 |
| DN | 0.91 | 0.08 | 1.00 | 0.97 | 0.43 | 0.23 | 2395.00 |
| GD T | 1.00 | 0.09 | 0.99 | 0.98 | 0.73 | 0.75 | 2279.00 |
| Treg | 1.00 | 0.06 | 1.00 | 0.99 | 0.64 | 0.67 | 1966.00 |
| Neg. sel. (2) | 1.00 | 0.07 | 0.99 | 0.98 | 0.64 | 0.67 | 1560.00 |
| Dying | 0.91 | 0.07 | 0.98 | 0.97 | 0.55 | 0.58 | 1552.00 |
| Neg. sel. (1) | 0.91 | 0.07 | 0.99 | 0.97 | 0.64 | 0.66 | 1206.00 |
| Mature cycling | 0.91 | 0.08 | 0.99 | 0.97 | 0.45 | 0.50 | 992.00 |
| Interferon sig. | 0.91 | 0.07 | 0.96 | 0.86 | 0.46 | 0.45 | 984.00 |
| NKT | 0.91 | 0.11 | 0.99 | 0.98 | 0.64 | 0.67 | 928.00 |
| Myeloid | 1.00 | 0.09 | 0.99 | 0.97 | 0.91 | 0.92 | 908.00 |
| Doublet | 0.64 | 0.11 | 0.81 | 0.84 | 0.94 | 0.76 | 677.00 |
| B | 0.91 | 0.30 | 0.98 | 0.92 | 0.82 | 0.81 | 106.00 |
| Erythrocyte | 0.64 | 0.41 | 0.94 | 0.85 | 0.49 | 0.62 | 43.00 |



</details>

# Model Properties

We provide here key parameters used to setup and train the model.

<details>
<summary><strong>Model Parameters</strong></summary>

These provide the settings to setup the original model:
```json
{
    "n_latent": 20,
    "gene_dispersion": "gene",
    "protein_dispersion": "protein",
    "gene_likelihood": "nb",
    "latent_distribution": "normal",
    "empirical_protein_background_prior": null,
    "override_missing_proteins": false
}
```

</details>

<details>
<summary><strong>Setup Data Arguments</strong></summary>

Arguments passed to setup_anndata of the original model:
```json
{
    "rna_layer": "counts",
    "protein_layer": null,
    "batch_key": "sample_id",
    "size_factor_key": null,
    "categorical_covariate_keys": null,
    "continuous_covariate_keys": null,
    "modalities": {
        "rna_layer": "rna",
        "protein_layer": "protein",
        "batch_key": "rna"
    }
}
```

</details>

<details>
<summary><strong>Data Registry</strong></summary>

Registry elements for AnnData manager:
| Registry Key             | scvi-tools Location                  |
|--------------------------|--------------------------------------|
| X                         | adata.mod['rna'].layers['counts']    |
| batch                     | adata.mod['rna'].obs['_scvi_batch']  |
| labels                    | adata.obs['_scvi_labels']            |
| latent_qzm                | adata.obsm['totalvi_latent_qzm']     |
| latent_qzv                | adata.obsm['totalvi_latent_qzv']     |
| minify_type               | adata.uns['_scvi_adata_minify_type'] |
| observed_lib_size         | adata.obs['observed_lib_size']       |
| proteins                  | adata.mod['protein'].X               |

- **Data is Minified**: False

</details>

<details>
<summary><strong>Summary Statistics</strong></summary>

| Summary Stat Key          | Value |
|--------------------------|-------|
| n_batch                   | 17 |
| n_cells                   | 72042 |
| n_extra_categorical_covs  | 0 |
| n_extra_continuous_covs   | 0 |
| n_labels                  | 1 |
| n_latent_qzm              | 20 |
| n_latent_qzv              | 20 |
| n_proteins                | 111 |
| n_vars                    | 4000 |

</details>


<details>
<summary><strong>Training</strong></summary>

<!-- If your model is not uploaded with any data (e.g., minified data) on the Model Hub, then make
sure to provide this field if you want users to be able to access your training data. See the
scvi-tools documentation for details. -->
**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/Thymus_CITE-seq/blob/main/totalVI_AllData/totalVI_thymus111.ipynb

</details>


# References

Steier, Z., Aylard, D.A., McIntyre, L.L. et al. Single-cell multiomic analysis of thymocyte development reveals drivers of CD4+ T cell and CD8+ T cell lineage commitment. Nat Immunol 24, 1579–1590 (2023). https://doi.org/10.1038/s41590-023-01584-0.