Ultra-fast single-cell workflow · powered by AltAnalyze3

Comprehensive online comparison of single-cell datasets in scALABLE

Aligns single-cell datasets to a reference atlas, provides an interactive Explore workspace for UMAP and gene-expression review, and supports group-based differential analysis in a dedicated Differential workspace. The app also performs cell-communication analysis, reporting ligand–receptor evidence per cell state.

Start an analysis Walkthrough video How to use

Analyses performed — select to enlarge

View at full resolution →

Accepted inputs

Upload one file per sample. Maximum 7 files per job, 1 GB per upload.

Format	Source
`.h5`	HDF5 count matrix from a standard droplet workflow
`.h5ad`	AnnData, as written by scanpy

Behaviour by input type

multiple .h5 files support group-based differential analysis
a single .h5ad also supports it when the .obs metadata contain multiple biological groups
a single .h5 upload does not enable group differential analysis

Compatible .obs metadata from uploaded .h5ad files are preserved and reused for Explore tab filtering, Differential cell-state selection and Differential biological-group selection.

Interface

Run

upload files
configure QC and alignment
review the reference preview before upload

Explore

inspect aligned UMAPs
inspect gene expression by UMAP or violin plot
review marker heatmaps and marker networks
run cell-communication analysis (ligand–receptor evidence per cell state)
download assignments, the combined h5ad and marker ZIP outputs

Differential

compare biological groups after alignment completes
inspect heatmap, volcano, network, GO terms and gene detail views

Alignment workflow

choose Species
choose Reference
add one or more samples
upload files
review QC settings
click Save QC and run

When alignment completes the app switches to Explore and results become available immediately. If the job supports grouped comparisons, the Differential tab becomes usable. Marker analysis (markerFinder, NetPerspective networks, heatmap PDF and TSV exports) is always run after alignment.

QC and alignment settings

Min genes	per-cell gene floor
Min counts	per-cell count floor
Min cells	per-gene cell floor
Mito %	mitochondrial ceiling
Minimum cosine similarity	alignment confidence floor
Ambient RNA correction	No / Yes

Ambient RNA correction is optional and applied per sample before alignment. In many standard droplet RNA datasets values near 20% are a reasonable expectation; higher contamination can occur in some assay types. Inspect your data and apply correction when needed.

References and optional modality imputation

Where a reference supports it, an Impute modality field appears in the QC step. The default is none.

Human reference	Imputable
LungMAP CellRef v1.1 (Guo 2023)	`lipids`
Lung HLCA (Sikkema 2023)	`lipids`
Lung ILD Atlas (Natri 2024)	`lipids`
BPD Atlas (Sun)	`lipids`
Bone marrow CITE-Seq (Zhang 2024)	`adt, metabolite, lipid, grn`

Mouse reference	Imputable
LungMAP CellRef v1.0	—
LungMAP Adult Lung/Airway v1.0	—
Lung Regeneration-Infection (Niethamer 2025)	—
Bone marrow CITE-Seq (Ferchen 2025)	`adt`

Method detail: cellHarmony technology and workflow · differential methods, defaults and filtering

Ready to run. Your data goes to the analysis tool, not to this site. Open scALABLE and start an analysis →