Skip to content

Quick Start

This guide walks through loading data and running a standard single-cell analysis in SCLab.

Load your data

SCLab accepts an AnnData object, a local file path, or a URL:

import scanpy as sc
from sclab import SCLabDashboard

adata = sc.read_10x_h5("filtered_feature_bc_matrix.h5")
dashboard = SCLabDashboard(adata, name="My Analysis")
display(dashboard)

from sclab import SCLabDashboard

dashboard = SCLabDashboard("path/to/data.h5ad", name="My Analysis")
display(dashboard)

from sclab import SCLabDashboard

dashboard = SCLabDashboard(
    "https://example.com/data.h5ad",
    name="My Analysis",
)
display(dashboard)

from sclab import SCLabDashboard

dashboard = SCLabDashboard()
display(dashboard)

SCLab will show a widget listing any AnnData objects already defined in your notebook, so you can pick one interactively.

Run the standard workflow

Once the dashboard is displayed, run each step from the left-hand panel in order:

Step What it does
QC Filter low-quality cells and genes, compute QC metrics
Preprocess Normalize, log-transform, and scale
PCA Reduce dimensions; inspect the variance ratio plot
Neighbors Build the k-NN graph used for UMAP and clustering
UMAP Compute a 2D embedding for visualization
Cluster Leiden clustering; results appear as leiden in the plot

Access components programmatically

The dashboard exposes its sub-components as properties:

# AnnData — all results are stored here
adata = dashboard.ds.adata

# Run a step from code
dashboard.pr.steps["QC"].run()

# Access the plotter
dashboard.pl

Next steps