Scvi Tools

scvi-tools — scVI / scANVI

scvi-tools (Gayoso et al. 2022, github.com/scverse/scvi-tools, BSD-3-Clause) wraps a family of deep generative models for single-cell omics. The scRNA-seq core is scVI (unsupervised batch-corrected latent embedding) and scANVI (scVI + a classifier head for semi-supervised cell-type label transfer). Both expect raw integer UMI counts and emit a low-dimensional X_scVI / X_scANVI that drops into the scanpy neighbors → leiden → umap pipeline.

How to run

scVI — batch-corrected latent space

import scanpy as sc
import scvi

adata = sc.read_h5ad("dataset.h5ad")
adata.layers["counts"] = adata.X.copy()          # preserve raw BEFORE any normalize/log1p
sc.pp.normalize_total(adata); sc.pp.log1p(adata) # optional, for HVG / plotting only
sc.pp.highly_variable_genes(adata, n_top_genes=2000, batch_key="batch", subset=True)

scvi.model.SCVI.setup_anndata(adata, layer="counts", batch_key="batch")
model = scvi.model.SCVI(adata, n_latent=30)
model.train(max_epochs=200, early_stopping=True, accelerator="gpu", devices=1)

adata.obsm["X_scVI"] = model.get_latent_representation()
adata.layers["scvi_normalized"] = model.get_normalized_expression(library_size=1e4)

scANVI — label transfer from a partially-annotated reference

lvae = scvi.model.SCANVI.from_scvi_model(
    model, labels_key="cell_type", unlabeled_category="Unknown",
)
lvae.train(max_epochs=20, n_samples_per_label=100, accelerator="gpu", devices=1)

adata.obsm["X_scANVI"] = lvae.get_latent_representation()
adata.obs["pred_cell_type"] = lvae.predict()

accelerator="gpu", devices=1 is the PyTorch-Lightning spelling; the legacy use_gpu= kwarg was removed in scvi-tools 1.x and now raises TypeError.

Differential expression

de = model.differential_expression(
    groupby="leiden", group1="3",   # group2=None → vs. all other cells
    mode="change", delta=0.25,
)
top = de.sort_values("proba_de", ascending=False).head(50)

For one-vs-rest leave group2 out — "rest" is scanpy's rank_genes_groups convention, not scvi-tools'; here group2 is a literal category name and "rest" would match zero cells.

scvi-tools ≥1.4 defaults to mode="vanilla", whose result columns are exactly:

['proba_m1', 'proba_m2', 'bayes_factor', 'scale1', 'scale2', 'raw_mean1',
 'raw_mean2', 'non_zeros_proportion1', 'non_zeros_proportion2',
 'raw_normalized_mean1', 'raw_normalized_mean2', 'comparison', 'group1',
 'group2']

— no lfc_*, no proba_de, no is_de_fdr_*. Pass mode="change" to get lfc_mean / lfc_median / proba_de / is_de_fdr_0.05. Sort on proba_de (or on bayes_factor if you deliberately stayed in vanilla mode).

Output format

Remote compute

A100-class GPU recommended for >50k cells. The prebuilt singlecell_gpu Modal env ships scvi-tools 1.4.2 + scanpy 1.11.5 + anndata 0.11.4 — read compute_details({provider: 'byoc:modal', mode: 'read'}) for the current image ref, then:

c = host.compute.create('byoc:modal', provider_params={'modal': {
    'image':  '<image ref from compute_details>',   # e.g. im-...
    'gpu':    'A100',
    'cpu':    8,
    'memory': 32768,
    'timeout': 3600,
}})
job = c.submit_job(
    intent="scVI+scANVI on 80k cells — 1×A100, ~15 min",
    inputs=[
        {"src": "dataset.h5ad", "dst_filename": "dataset.h5ad"},
        {"src": "pipeline.py",  "dst_filename": "pipeline.py"},
    ],
    command="python pipeline.py",
    outputs=["out/**"],
    timeout_seconds=2400,
)
print(job.job_id)   # cell ends here — kernel never blocks on compute

h5ad_safe_obs is auto-loaded into the local analysis kernel only — in pipeline.py running on Modal, paste the helper at the top of the script (or inline the pd.Index(np.asarray(..., dtype=object)) coercion) before .write_h5ad().

Then call the wait_for_notification brain-tool. When compute_done arrives, save_artifacts(payload["featured_files"]). For the full result dict, re-enter the kernel and bind the compute handle (not the job) separately — .close() lives on the handle, not on the job:

h = host.compute.create('byoc:modal')
res = h.attach_job(job_id).result()   # output_files, remote_workdir, ...
h.close()

See the remote-compute-modal skill for orchestration details.

Gotchas

Troubleshooting

Next: cluster on X_scVI with scanpy (sc.pp.neighbors(use_rep="X_scVI") → sc.tl.leiden → sc.tl.umap); for spatial deconvolution train cell2location / DestVI / Tangram on the scRNA-seq reference.