An individual trait is a single trait dataset with its own accession or identifier. A meta trait combines related individual traits under one trait code. Trait refers to either level. This organization does not by itself imply a new statistical meta-analysis.
The cell annotation directory provides readable pages for every cell type, subtype and state. Each page contains the same marker genes, annotation genes, biological rationale and functional implications shown in the Cell Browser, plus the full-atlas cell count and related populations. No interactive viewer or large data download is needed to read these pages.
The Evidence reports directory provides complete, paginated meta-trait, individual-trait and gene-expression rankings for each cell type, subtype and state, and cell-population rankings for each meta trait or individual trait. It also includes the available DE gene and MAGMA gene-association exports, whose upstream limits are explained on each page. Follow the page links to read the full available lists rather than only the highest-scoring entries. These ordinary HTML pages can be read without running Cell Browser or downloading a scientific matrix.
Scope-specific evidence reports provide the complete evidence used by the Cell Browser’s AI ranking buttons, including prepared example views.
Rankings describe associations, not proof of causality. Each ranking page identifies its scope and score and links to related annotations and the interactive Cell Browser for further exploration.
The SCREATE Genome Browser uses JBrowse to connect human genomic regions with genes, transcriptional cis-regulatory elements (tCREs), promoter–distal links, variants and trait associations. Open a trait or gene view to explore its relevant tracks and associated-feature table.
Select a trait in the Trait Catalogue or Trait UMAP, then use the Genome Browser button. The landing page examples pair Cell Browser and Genome Browser views. The middle table provides links between gene and trait views; pane controls adjust the space available for tracks and the table.
Click a gene or trait name in the middle table to highlight its linked features in red and jump to that row’s region in the existing genome pane. Trait selection uses the exact meta-trait code or individual-trait ID and filters the table to its matching rows. Use Clear selection to restore the full table. The small arrow beside the name opens the corresponding gene or trait view in a new tab.
The sparkle AI icon after a gene or trait’s arrow opens Google AI Mode with a prompt for that gene–trait pair. It summarizes the clicked row and related rows in the current filtered table across all pages, using the available fine-mapping, regulatory and cellular evidence. It asks for biological interpretation, caveats, potential druggability and sources. The summary is sent to Google only when you click the icon; no AI request is made automatically.
These prompts define the ICE-CREAM (IC) score: a cell/meta-cell trait-enrichment score, reported as −log10(P), comparing trait-heritability-weighted tCRE-module usage against an expression-matched permutation background. They explain that IC-score correlations compare tCRE expression/activity with the selected trait’s IC score across matched meta-cells. The current tracks use weighted correlations in L1S scope (between cell types) and L2 scope (within cell types). A correlation coefficient is not itself an IC score or proof of causality.
The Cell Browser is an interactive UMAP viewer for exploring cell types, subtypes, cell states, genes, and traits. Use it to move from a broad overview of the atlas to more specific cell populations, and to visualize where genes or traits are enriched.
The left panel can show cell annotations at different resolutions:
Click a cell group name to highlight it on the UMAP. Double-click Type or Subtype entries to drill down into that population.
The Features panel contains three tabs:
Select a trait or gene to color the UMAP by its score or expression. When a cell group is highlighted, traits and genes are ranked by their average value in that selected group.
In the Traits tab, the Enrichment button switches trait display to CSEA normalized enrichment scores (NES). With Enrichment on, trait coloring and trait rankings use NES values for the currently selected cell-group resolution: Type uses L1 NES, Subtype uses L1S NES, and State uses L2 NES. With Enrichment off, the browser uses the default trait-associated signal.
Plain-click a trait to select and color by that single trait. To compare two traits or two genes,
first click the feature that should be the reference, then Shift-click the second feature.
The browser colors cells by the normalized difference
first feature - second feature and ranks cell groups by that difference score.
This is useful for asking which cell types are more strongly associated with one trait than another.
To average multiple traits, Ctrl-click on Windows or Linux, or Cmd-click on macOS, to add or remove each trait. In regular trait mode this averages their raw scores; with Enrichment on, it averages their NES values instead. Opening the Cell Browser from a selected trait code with multiple summary-statistic IDs on the main Traits page automatically averages all summary statistics in that trait code.
age onset -first.The information panels show details for the selected cell group, scope, gene, or trait. Use the pin button to keep the current information visible while browsing other items.
The Ask AI button opens a new Google AI/search tab with a ready-made prompt based on the current selection. The browser includes the top 10 ranked items and their scores so that the answer can distinguish between strong and weak enrichments.
No AI request is sent automatically; the query is only opened when the button is clicked.
The browser updates the URL as you navigate. You can copy the URL to share the current view, including the selected gene, trait, feature difference, multi-trait average, or multi-gene average when one is active.
Example URL parameters used by the browser:
?view=overview: open the full atlas.?view=l1&name=...: open a Type-level view.?view=l1s&parent=...&base=...: open a Subtype-level view.?view=lx&name=...: open an LX scope.&selk=gene&selname=EBF1: color by a selected gene.&selk=trait&selname=...: color by a selected trait.&selk=trait&selname=...&selname2=...: color by the difference between two traits.&selk=trait_avg&selname=<semicolon-separated IDs/names>: color by a manually selected trait average.&selk=trait_avg&selcode=<trait-code>&sellabel=<optional label>: color by the average of every summary statistic in a grouped trait code.&selk=gene&selname=...&selname2=...: color by the difference between two genes.&selk=gene_avg&selname=EBF1;MS4A1: color by the average of multiple genes.&nes=1: keep trait Enrichment / CSEA NES mode enabled when sharing the view.