HyCell report benchmark on seven held-out atlases
Scores frozen single-cell embeddings on seven atlases excluded from pretraining, for biological conservation, batch correction and annotation transfer.
Results are recorded on the linked protocols, tasks or configurations listed below. Their settings and evaluations are kept separate.
Overview
An explanatory profile has not yet been reviewed for this record.
Consult the linked sources for architecture or protocol details. Missing evidence is not evidence of a missing capability.
Results on linked records
This benchmark has no results of its own in this release. 16 linked records have results. They are not assigned to the underlying benchmark: each record keeps its own settings and evaluations, and its scores are on its own page.
- Annotation of a held-out batch after fine-tuning (HyCell report) Protocol · part of this benchmark · 4 evaluations · 4 results
- Annotation transfer with 15-nearest neighbours (HyCell report) Protocol · part of this benchmark · 16 evaluations · 16 results
- Annotation transfer with logistic regression (HyCell report) Protocol · part of this benchmark · 16 evaluations · 16 results
- scIB batch correction (HyCell report) Protocol · part of this benchmark · 14 evaluations · 14 results
- scIB biological conservation (HyCell report) Protocol · part of this benchmark · 55 evaluations · 55 results
- Naive against effector CD4 T-cell split, ARI (HyCell report) Protocol · part of this benchmark · 1 evaluation · 1 result
- Naive against effector CD4 T-cell split within donors, ARI (HyCell report) Protocol · part of this benchmark · 3 evaluations · 3 results
- CD4 T-cell split against sequencing chemistry, NMI (HyCell report) Protocol · part of this benchmark · 1 evaluation · 1 result
- Composite score S (HyCell report) Protocol · part of this benchmark · 96 evaluations · 96 results
- Decoded against counted fold change, plasma against naive B (HyCell report) Protocol · part of this benchmark · 1 evaluation · 1 result
- Gene-program consistency across donors, naive to effector CD4 T cells (HyCell report) Protocol · part of this benchmark · 2 evaluations · 2 results
- Gene-program consistency across donors, classical to nonclassical monocytes (HyCell report) Protocol · part of this benchmark · 2 evaluations · 2 results
- Gene-program consistency across donors, naive B to plasma cells (HyCell report) Protocol · part of this benchmark · 2 evaluations · 2 results
- gamma-delta T against tissue-resident CD8 T split, ARI (HyCell report) Protocol · part of this benchmark · 2 evaluations · 2 results
- B-cell to plasma-cell pseudotime ordering, all cells (HyCell report) Protocol · part of this benchmark · 4 evaluations · 4 results
- B-cell to plasma-cell pseudotime ordering, donor mean (HyCell report) Protocol · part of this benchmark · 4 evaluations · 4 results
Methods and evaluation design
Procedure, tasks and evaluated configurations
Additional methodological details, tasks and evaluated configurations have not yet been extracted into a structured profile for this benchmark. Consult the linked sources and evaluation records below.
Baseline coverage
Reference methods help show what a model adds beyond simple controls. We track a null control and a conventional method for each protocol.
0 of 32 active baseline roles have published Rewire measurements in this release. Measurements on a selected protocol do not establish coverage of an entire suite.
Baseline status by linked protocol
- Annotation of a held-out batch after fine-tuning (HyCell report) · 0/2 roles measured
- Annotation transfer with 15-nearest neighbours (HyCell report) · 0/2 roles measured
- Annotation transfer with logistic regression (HyCell report) · 0/2 roles measured
- scIB batch correction (HyCell report) · 0/2 roles measured
- scIB biological conservation (HyCell report) · 0/2 roles measured
- Naive against effector CD4 T-cell split, ARI (HyCell report) · 0/2 roles measured
- Naive against effector CD4 T-cell split within donors, ARI (HyCell report) · 0/2 roles measured
- CD4 T-cell split against sequencing chemistry, NMI (HyCell report) · 0/2 roles measured
- Composite score S (HyCell report) · 0/2 roles measured
- Decoded against counted fold change, plasma against naive B (HyCell report) · 0/2 roles measured
- Gene-program consistency across donors, naive to effector CD4 T cells (HyCell report) · 0/2 roles measured
- Gene-program consistency across donors, classical to nonclassical monocytes (HyCell report) · 0/2 roles measured
- Gene-program consistency across donors, naive B to plasma cells (HyCell report) · 0/2 roles measured
- gamma-delta T against tissue-resident CD8 T split, ARI (HyCell report) · 0/2 roles measured
- B-cell to plasma-cell pseudotime ordering, all cells (HyCell report) · 0/2 roles measured
- B-cell to plasma-cell pseudotime ordering, donor mean (HyCell report) · 0/2 roles measured
Protocol coverage CSV (gzip) · Model evaluation matrix (gzip) · Source table (gzip) · Release and checksums (gzip)
Coverage is derived from release 2026-11-21-c8bea25e5abb. Source citations describe the original records; they do not validate an unreviewed baseline proposal. No results have been generated by this audit.
Run instructions
No runnable recipe has been reviewed for this benchmark. Dataset access, model requirements, licences and compute requirements must be checked against its sources before execution.
Evidence
Source checking verifies the cited claim or transcription. It does not establish independent reproduction.
Evidence table
Inspect claims, sources and review details
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
0 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|
No evidence rows match these filters. Choose another scope or clear the search.
Sources and history
Release 2026-11-21-c8bea25e5abb · Record review: source checked
1 source record and release history
- HyCell: a Compact Embedding Model for Single-Cell Transcriptomes (Hyra Team, 2026-11-01) · Original source · Dated 2026-11-01 on page 1; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (9 pages)
Technical metadata and record identifiers
Technical metadata for this record is tracked under stable ID hycell-20261106-benchmark-held-out-atlases.
Source locator: Section 3.1; Tables 2 to 4
Related records
- evaluates task: Donor-held-out reference mapping
- subject: HyCell report: bio first atlases
- subject: HyCell report: cd4 chemistry range
- subject: HyCell report: cd4 others bound
- subject: HyCell report: cell level scope
- subject: HyCell report: composite weights
- subject: HyCell report: excluded atlases
- subject: HyCell report: finetune knn batches
- subject: HyCell report: finetune lr batches
- subject: HyCell report: gd published bound
- subject: HyCell report: headline level
- subject: HyCell report: knn gap
- subject: HyCell report: label trained
- subject: HyCell report: lr gap
- subject: HyCell report: median pt gc
- subject: HyCell report: median pt naive
- subject: HyCell report: median pt plasma
- subject: HyCell report: median pt plasmablast
- subject: HyCell report: pca condition
- subject: HyCell report: pseudotime vs pca
- subject: HyCell report: readme best s
- subject: HyCell report: size range
- subject: HyCell report: stack condition
- part of: Annotation of a held-out batch after fine-tuning (HyCell report)
- part of: Annotation transfer with 15-nearest neighbours (HyCell report)
- part of: Annotation transfer with logistic regression (HyCell report)
- part of: scIB batch correction (HyCell report)
- part of: scIB biological conservation (HyCell report)
- part of: Naive against effector CD4 T-cell split, ARI (HyCell report)
- part of: Naive against effector CD4 T-cell split within donors, ARI (HyCell report)
- part of: CD4 T-cell split against sequencing chemistry, NMI (HyCell report)
- part of: Composite score S (HyCell report)
- part of: Decoded against counted fold change, plasma against naive B (HyCell report)
- part of: Gene-program consistency across donors, naive to effector CD4 T cells (HyCell report)
- part of: Gene-program consistency across donors, classical to nonclassical monocytes (HyCell report)
- part of: Gene-program consistency across donors, naive B to plasma cells (HyCell report)
- part of: gamma-delta T against tissue-resident CD8 T split, ARI (HyCell report)
- part of: B-cell to plasma-cell pseudotime ordering, all cells (HyCell report)
- part of: B-cell to plasma-cell pseudotime ordering, donor mean (HyCell report)