Guide

How to Interpret Scion

Use this page when you need the meaning of a score, badge, or status label. For general orientation, start from the landing page. For project context and source links, use About.

Comparability Score

This is a ranking aid for likely overlap, not a claim that two studies are equivalent. It rewards shared biology and shared technical structure, then compresses that into a quick reading. A high score means “look here first,” not “pool these records without caveats.”

Current heuristic:
+25 if all selected datasets share the same cell type
+10 if all selected datasets share the same species
+5 each for shared organelle pairs, capped at +20
+3 each for shared metric families, capped at +15
+10 if all selected datasets share the same modality family
+10 if they share a comparator class
+10 if every selected record has metadata completeness of at least 0.8
The total is capped at 100.
Current reading bands: 75+ high overlap, 50-74 moderate overlap, and below 50 analog-level comparison.

Metadata Completeness

This measures how fully a record reports standardized fields such as modality, resolution, sample size, and curation status. It is not a scientific-quality score.

+20% identity fields present in the seed record
+15% biological context present
+15% resolution reported, or +5% when not reported
+20% organelles captured are listed
+10% metric families are listed
+10% sample size is reported
+10% public data is known, or +5% when no public data source is known

Public Data Status

None: no reusable public data source is known from the current corpus materials.

Partial: some public underlying data or assets are available, but not necessarily the full dataset.

Complete: reusable public data is known to exist for the dataset in a stronger form.

Included vs Borderline

Included records met the primary corpus criteria. Borderline records are still useful, but they usually have a methodological, reporting, or whole-cell-coverage caveat that should remain visible.

Badge Legend

ResResolution is explicitly reported.
SSSample size is explicitly reported.
BoundaryWhole-cell boundary confirmation is present.
Public DataReusable public data is known to exist.
BorderlineUseful near-miss record; keep it in context, but verify more carefully.

Compare Workflow

The current compare workflow starts from the corpus. Add two or more datasets from the table or card view, then use the compare drawer or Compare link. The empty compare page is just a destination, not where selection begins.

Compare is dataset-level. Some papers contribute multiple records, so same-paper comparisons can be useful for checking conditions or modalities but should not be read as independent cross-study validation.

Best Next Step

Open the CorpusOpen About