Step-by-step guides for exploring species, searching genes, comparing across species,
interpreting correlations, and downloading data. Each workflow runs independently — jump to any of them.
⚠️ The walkthroughs use SP121 (mouse), the site’s demo species. All data shown are real and served
from the same atlas packages you can download; SP121 is simply the species preloaded in the Demo menu.
01
Explore Species Information
⏲ ~5 min
See which species are covered and how each atlas is organized by system, tissue and cell type.
Open the Species Atlas from the top navigation “Longevity → Atlas”.
You will see: The analysis module for the selected tissue context: UMAP canvas plus result panels. A “How is this computed?” chip links to the matching Methods section.
02
Search Genes
⏲ ~4 min
Find a gene and inspect its expression across the atlas.
You will see: Cell-type level correlations for the selected context. SP121 also has a downloadable cell-level table (cell_level_celltype_correlation.csv, linked on the module page).
Before quoting any number, read the Methods section for this module.
You will see: Correlations use Spearman’s rank correlation by default; multiple-testing is controlled with Benjamini–Hochberg FDR; cell-level notes explain aggregation. Correlation here is association, not causation.
05
Data Download
⏲ ~4 min
Download count matrices, atlas packages and cell-level metadata for offline analysis.
You will see: Real, server-verified files: 173 CancerSCEM count matrices, bulk matrices, four curated h5ad datasets, and a filterable table of 141 species atlas packages.
In the species table, use “Atlas package” for the full bundle or “Cell-level metadata” for a per-cell CSV.
You will see: A generated download: the atlas package (tar.gz with embedding + legends + config) or cell_idx,x,y,cell_type CSV, generated on demand and cached.
To cite or reproduce a specific site state, check the archived snapshots.
An export link takes a few seconds to respond — is it broken?
No. Atlas packages and cell-level CSVs are generated on first request and then cached. The first click on a large species can take a few seconds; later downloads are instant.
A correlation or comparison page looks empty for my gene.
Cross-species views are restricted to genes with one-to-one orthologs, and correlation views need the gene to be present in the selected dataset. Try a well-studied gene (e.g. Mtor, Sirt1, Igf1r) or another species.
I find dark pages hard to read — is there a light mode?
Yes. Click the Light button at the right end of the top navigation to switch the site to a high-contrast light theme (WCAG AA-recalibrated palette, larger body text and looser line spacing on content pages). Your choice is remembered on this device, and on a first visit the site follows your operating system’s light/dark preference.
How do I cite a number from the site?
Check Methods & Data Processing for how the value is computed, and the Versions page to identify the current site release. Correlations are associations — not causal claims.