ScarfWeb is the browser-based analysis layer for single-cell omics. Explore clusters, run marker analysis, annotate cell types, and share reproducible findings — without writing code or provisioning compute.
New to ScarfWeb?Upload your own data or connect to curated datasets and start working immediately. ScarfWeb handles the infrastructure so your team can focus on the biology. Publication-ready outputs, reproducible workflows, and real-time collaboration are built in.
From data upload to annotated clusters and shareable figures.
Every step from QC to differential expression runs through a visual interface with adjustable parameters. Researchers drive their own analysis without writing code or waiting on bioinformatics queues. The same workflows are fully reproducible, versioned, and auditable.
CyteType runs directly inside your analysis session. Automated cell annotation with confidence scoring, marker-level evidence, ontology mapping, and literature linking. Review and override at the cluster level. No separate tool, no export required.
Bench scientists and bioinformaticians work in the same dataset, the same session. Share results via public links or private access, publish to Nygen Database, or export for scripted pipelines. The analysis stays in one place regardless of who needs to see it.
Every step of single-cell analysis — data management, processing, visualization, statistical testing, and collaboration — in one browser-based workspace.
Get your data in. Keep it organized.
Upload preprocessed count matrices in common formats including MTX, H5AD, H5, and CSV. Drag, drop, analyze.
Upload transcriptomic and protein-level data together. ADT and HTO data are detected and separated automatically during import.
Combine datasets across samples, batches, or conditions. Demultiplex HTO-labeled pools without external tools.
Bring in sample annotations, condition labels, or existing clustering results from Seurat, Scanpy, or other pipelines. No reformatting required.
Update cell-level and sample-level annotations across your dataset in one step.
From raw counts to clusters. Fully parameterized, zero code.
Filter by detected genes, total UMIs, mitochondrial content, and ribosomal content. Thresholds are adjustable and visualized in real time.
Custom block-listing, variance and mean filters, and minimum cell-count thresholds. Control exactly which genes drive your dimensionality reduction.
Harmonize datasets from different experiments or sites before joint analysis. Remove technical variation, preserve biological signal.
PCA for computation, UMAP and t-SNE for visualization. Select principal components and adjust parameters directly in the interface.
Leiden and Paris algorithms with adjustable resolution. Fine or coarse, depending on what your biology requires.
Every parameter set and result is saved as a versioned run. Compare across runs, roll back, or branch your analysis at any point.
Interrogate your data in the browser. No notebooks, no waiting.
Pan, zoom, select, and color cells by cluster, gene expression, metadata, or custom categories. All rendering happens in real time.
Define cell populations using any combination of cluster identity, metadata, and expression thresholds. Save them for downstream comparison.
Visualize any gene's expression directly on the embedding. Search, overlay, compare.
Create, save, and score custom gene sets. View aggregate signature expression on scatter plots, dot plots, and feature plots.
AI-driven cell type annotation with confidence scores. Review, override, or accept at the cluster level.
Go deeper. Ask harder questions of your data.
Compare any two or more cell selections. Results include fold change, p-values, and volcano plot visualization. Design complex comparisons across conditions, clusters, or custom groups.
Quantify and test changes in cell type proportions across conditions. Identify shifts that are statistically significant, not just visually suggestive.
Order cells along a biological process. Identify gene modules that change across pseudotime and visualize them as expression heatmaps.
Automatic one-vs-all marker detection for every cluster. Sortable by marker score, fold change, and fraction expressed. One-click violin and dot plot generation.
Share evidence, not screenshots.
Generate a shareable, read-only link to any analyzed dataset. Reviewers and collaborators can explore your data interactively without an account.
Share datasets privately with specific collaborators via email. Control who sees what.
Make your analysis publicly available on Nygen Database in a few clicks. No web server setup, no hosting overhead.
Group datasets, analyses, and annotations into projects. Manage the full lifecycle of a study in one workspace.
Every layer of ScarfWeb is built to keep your data secure, your analysis reproducible, and your outputs defensible.
Runs entirely in the browser on scalable cloud infrastructure across 22 AWS regions. Upload your data and start working — no local compute, no pipeline setup.
Peer-reviewable figures, exportable tables, and auto-generated methods sections. Every output traces back to the exact parameters used.
ISO 27001 certified, SOC 2 compliant, fully GDPR adherent. Data encrypted at rest and in transit with comprehensive backups.
Full version control across every analysis session. Every parameter, filter, and selection logged and shareable.
Start analyzing in the browser today, or talk to our team about how ScarfWeb fits your workflow.
New to ScarfWeb?