What you need
A 10x Genomics output folder (matrix.mtx, features.tsv, barcodes.tsv, gzipped or not), an exported Seurat or h5ad object, or a counts table, plus cell or sample metadata if you have it.
Step by step in RNA Atlas
- Try the example first. RNA Atlas opens with a synthetic single-cell dataset and a guided example, so you can see every step before using your data.
- Import your data. Choose ＋ Import dataset. The importer previews files, detects the format and checks gene identifiers. Projects are saved on this device and can be exported.
- Quality control. Set thresholds for detected genes, library size and mitochondrial fraction and see how many cells each removes before you apply them.
- Embeddings and clusters. Compute PCA and UMAP in the browser; colour cells by cluster, sample, condition or any gene.
- Explore genes. In Gene explorer, plot a gene by cluster or condition, or a whole panel as a bubble plot or heatmap. Find papers searches the literature for it.
- Compare conditions correctly. In Differential analysis, cells are summed to pseudobulk per biological sample before the contrast, so cells are not treated as independent replicates.
- Pathways and export. Run pathway enrichment with bundled Reactome sets, then export figures, tables, the exact settings and a methods text.
Good practice the tool builds in
- Biological samples, not cells, are the unit of replication for condition comparisons.
- Every filter and parameter is recorded with the result.
- Large datasets can use an optional engine you run on your own computer (Scanpy, Harmony, Scrublet, Leiden), with explicit consent for each transfer.
Open RNA Atlas How it is validated
Questions
Is it really free and private?
Yes. Analyses run in your browser; nothing is uploaded to a SolvesEverything server and no account is needed.
How large a dataset can it handle?
Browser memory sets the limit; tens of thousands of cells are practical on a laptop. The optional local engine handles larger datasets.
Can I reproduce the analysis in R?
Yes. Exports include settings, methods text and, for model fits, exact R code.
Found something wrong or unclear? Write to contact@solveseverything.com. See also: how results are validated.