Skip to content

UCE

✅ Status: ready | Version: 4-layer

Overview

Broadest species support (7 species), 1280-dim embeddings, universal cell embedding via protein structure

When to choose UCE

User has non-human/non-mouse species (zebrafish, frog, pig, macaque, lemur), or needs cross-species comparison

Specifications

Property Value
Model UCE
Version 4-layer
Tasks embed, integrate
Modalities RNA
Species human, mouse, zebrafish, mouse_lemur, macaque, frog, pig
Gene IDs symbol
Embedding Dim 1280
GPU Required Yes
Min VRAM 16 GB
Recommended VRAM 16 GB
CPU Fallback No
Adapter Status ✅ ready

Quick Start

import omicverse as ov

# 1. Check model spec
info = ov.fm.describe_model("uce")

# 2. Profile your data
profile = ov.fm.profile_data("your_data.h5ad")

# 3. Validate compatibility
check = ov.fm.preprocess_validate("your_data.h5ad", "uce", "embed")

# 4. Run inference
result = ov.fm.run(
    task="embed",
    model_name="uce",
    adata_path="your_data.h5ad",
    output_path="output_uce.h5ad",
    device="auto",
)

# 5. Interpret results
metrics = ov.fm.interpret_results("output_uce.h5ad", task="embed")

Input Requirements

Requirement Detail
Gene ID scheme symbol
Preprocessing Standard log-normalization. Model handles tokenization internally.
Data format AnnData (.h5ad)
Batch key .obs column for batch integration (optional)

Output Keys

After running ov.fm.run(), results are stored in the AnnData object:

Key Location Description
X_uce adata.obsm Cell embeddings (1280-dim) 
import scanpy as sc

adata = sc.read_h5ad("output_uce.h5ad")
embeddings = adata.obsm["X_uce"]  # shape: (n_cells, 1280)

# Downstream analysis
sc.pp.neighbors(adata, use_rep="X_uce")
sc.tl.umap(adata)
sc.tl.leiden(adata, resolution=0.5)
sc.pl.umap(adata, color=["leiden"])

Resources

Hands-On Tutorial

For a step-by-step walkthrough with code, see the UCE Tutorial Notebook.