Backport PR #3575 on branch 1.4.x (Model: DiagVI (new)) (#3810)

Backport PR #3575: Model: DiagVI (new)

Co-authored-by: Ori Kronfeld <ori.kronfeld@weizmann.ac.il>

Author: meeseeksmachine

.github/workflows/test_linux_nonjax.yml

@@ -58,7 +58,7 @@ jobs:
           python -m uv pip install --system "coverage"
           python -m uv pip install --system "igraph"
           python -m uv pip install --system "leidenalg"
-          python -m uv pip install --system "scvi-tools[autotune,hub,mlflow,file_sharing,regseq,parallel,interpretability] @ ."
+          python -m uv pip install --system "scvi-tools[autotune,hub,mlflow,file_sharing,regseq,parallel,interpretability,diagvi] @ ."
 
       - name: Run pytest
         env:

CHANGELOG.md

@@ -11,6 +11,7 @@ to [Semantic Versioning]. The full commit history is available in the [commit lo
 
 - Add support for Python 3.14, {pr}`3563`.
 - Add support for Pandas3, {pr}`3638`.
+- Add {class}`scvi.external.DIAGVI` for integrating unpaired single-cell datasets, {pr}`3575`.
 - Add MuData support to {class}`scvi.external.TOTALANVI` {pr}`3797`.
 
 #### Fixed

docs/api/developer.md

@@ -102,6 +102,10 @@ Parameterizable probability distributions.
    distributions.ZeroInflatedNegativeBinomial
    distributions.JaxNegativeBinomialMeanDisp
    distributions.BetaBinomial
+   distributions.Normal
+   distributions.Log1pNormal
+   distributions.ZeroInflatedLogNormal
+   distributions.ZeroInflatedGamma
 
 ```
 
@@ -197,6 +201,7 @@ Module classes in the external API with respective generative and inference proc
    external.scviva.nicheVAE
    external.scviva.NicheLossOutput
    external.sysvi.SysVAE
+   external.diagvi.DIAGVAE
 
 ```
 

docs/api/user.md

@@ -70,6 +70,7 @@ import scvi
    external.RESOLVI
    external.SysVI
    external.SCVIVA
+   external.DIAGVI
 ```
 
 ## Data loading

docs/tutorials/index_multimodal.md

@@ -8,6 +8,8 @@ notebooks/multimodal/scarches_scvi_tools
 notebooks/multimodal/totalVI_reference_mapping
 notebooks/multimodal/cite_scrna_integration_w_totalVI
 notebooks/multimodal/MultiVI_tutorial
+notebooks/multimodal/DiagVI_spatial_proteomics.ipynb
+notebooks/multimodal/DiagVI_spatial_transcriptomics.ipynb
 ```
 
 ```{customcard}
@@ -45,3 +47,17 @@ Use totalVI to integrate CITE-seq and scRNA-seq datasets
 
 Go through the MultiVI workflow to perform joint analysis of paired and unpaired multi omic data
 ```
+
+```{customcard}
+:path: notebooks/multimodal/DiagVI_spatial_proteomics
+:tags: Analysis, Integration, Modality-imputation, Dimensionality-reduction
+
+Perform integration of spatial proteomics and single-cell transcriptomics data with DiagVI
+```
+
+```{customcard}
+:path: notebooks/multimodal/DiagVI_spatial_transcriptomics
+:tags: Analysis, Integration, Modality-imputation, Dimensionality-reduction
+
+Perform integration of spatial and single-cell transcriptomics data with DiagVI
+```

docs/user_guide/index.md

@@ -143,6 +143,9 @@ scvi-tools is composed of models that can perform one or many analysis tasks. In
    * - :doc:`/user_guide/models/multivi`
      - Integration of paired/unpaired multiome data, missing modality imputation, normalization of other cell- and sample-level confounding factors
      - :cite:p:`AshuachGabitto21`
+   * - :doc:`/user_guide/models/diagvi`
+     - Diagonal integration of unpaired multiome data, dimensionality reduction, cross-modality imputation, cell label transfer
+     - []
 
 ```