rewire.it
Benchmark

SegmentNT human genome annotation

Fourteen genomic-element annotation tasks from the SegmentNT study, with complete per-nucleotide MCC and auPRC comparisons from Supplementary Tables 2 and 3. Each element and metric remains a separate comparison.

294 evaluations · 588 metric rows

Overview

Fourteen genomic-element annotation tasks from the SegmentNT study, with complete per-nucleotide MCC and auPRC comparisons from Supplementary Tables 2 and 3. Each element and metric remains a separate comparison.

Consult the linked sources for architecture or protocol details. Missing evidence is not evidence of a missing capability.

Results

Each comparison retains its reviewed evaluation scope, dataset and metric. Results are shown without a pooled ranking.

SegmentNT human genome annotation 3UTR auPRC: 3UTR: per-nucleotide annotation (auPRC)

auprc (dimensionless) · Higher values are better.

Every method SegmentNT human genome annotation reports on 3UTR: per-nucleotide annotation (auPRC), scored with Area under precision-recall curve on Human genome 3UTR test chromosomes 20 and 21.

SegmentNT human genome annotation 3UTR auPRC: 3UTR: per-nucleotide annotation (auPRC) · Human genome 3UTR test chromosomes 20 and 21 (SegmentNT human genome annotation split)

Evidence origin: Author-reported evaluation. Numerical source review does not establish independent reproduction.

SegmentNT supplementary information: complete Tables 2 and 3 · Supplementary Table 3, 3UTR column; primary article Sec16–17

Ten test-set samplings with sliding windows beginning at different genomic starting positions. Mean plus reported standard deviation; not ten training seeds.

All comparison limitations (5)
  • Ten test-set samplings with sliding windows beginning at different genomic starting positions. Mean plus reported standard deviation; not ten training seeds.
  • Table names the v1 backbone NTv1 human; 2.5B; Sec17 describes NT v1 2.5B 1000G. Preserve exact table label and methods context.
  • Source excludes homologous test genes but not homologous distal regulatory elements, which may inflate those region scores.
  • Input lengths and training setups differ. Architecture baselines are newly trained from random initialization, not original pretrained checkpoint results.
  • No aggregation across genomic elements or metrics. Split manifests and checkpoint revisions remain unextracted.

Automated source review: 2026-09-23.

No unavailable values; missing scores remain labelled and are never plotted as zero.

Showing 12 of 21 matching rows.

Tested configuration
00.2510.5020.7521
Reported score
  1. SegmentBorzoi-524kb0.90 (± 0.103)
  2. SegmentNT-20kb0.70 (± 0.008)
  3. SegmentNT-30kb0.70 (± 0.006)
  4. SegmentEnformer-30kb0.67 (± 0.011)
  5. SegmentBorzoi-30kb0.67 (± 0.010)
  6. SegmentEnformer-196kb0.65 (± 0.051)
  7. SegmentNT-10kb0.64 (± 0.008)
  8. SegmentNT-3kb-single-task (x14)0.46 (± 0.005)
  9. SegmentNT-3kb0.43 (± 0.010)
  10. SegmentNT-3kb (NTv1 human; 2.5B)0.36 (± 0.014)
  11. SegmentNT-3kb (only head)0.29 (± 0.003)
  12. SpliceAI arch.0.23 (± 0.006)

Dots show point estimates. Whiskers show only explicitly defined uncertainty (standard deviation, standard error or a labelled interval); their definitions remain in Table. Unresolved uncertainty is not plotted. Differences do not establish statistical significance.

Methods and evaluation design

Procedure, tasks and evaluated configurations

Evaluation design

Benchmarks bring together tasks and protocols. A task describes the biological question; a protocol defines a particular test.

Protocols

These source-backed links do not make different protocols or scores interchangeable.

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 56 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

Protocol coverage CSV · Model evaluation matrix · Source table · Release and checksums

Coverage is derived from release 2026-09-23-2b89723c6dd9. Source citations describe the original records; they do not validate an unreviewed baseline proposal. No results have been generated by this audit.

Run this benchmark

Choose a concrete protocol before running an evaluation. Its inputs, split and scoring rules determine which results can be compared.

Run this benchmark

Test chromosomes 20 and 21; validation chromosome 22; training remaining chromosomes. Test chunks with genes homologous to train/validation genes excluded using Ensembl BioMart accessed 2024-05-08; homologous distal regulatory elements not excluded. Rewire has not executed this evaluation; exact sequence manifests and checkpoint hashes remain unextracted.

A maintained rewire runner has not been verified for this benchmark. Check data access, weights, licences, dependencies and hardware in the linked official documentation; requirements have not been fully extracted.

SegmentNT supplementary information: complete Tables 2 and 3; segmentnt: Journal full-text XML · Primary article Sec16–17
Strengths, limitations and unresolved questions

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

Claims, original sources and review scope · Release 2026-09-23-2b89723c6dd9
Property and statementOriginal source and locationReview and provenance

No evidence rows match these filters. Choose another scope or clear the search.

All evidence (CSV)All evidence (JSONL)How to read the evidence

Sources and history

View linked audit checks and correction history

Release 2026-09-23-2b89723c6dd9 · Record review: source checked

2 source records and release historyDownload this release
Technical metadata and extraction receipts

Stable ID: segmentnt-human-genome-annotation-2025

areas
genomics
source locator
Supplementary Tables 2 and 3 (PDF pages 4 and 5); primary article Methods / Model training and evaluation (Sec16), Model ablations and baselines (Sec17)
run documentation
record id: segmentnt-human-genome-annotation-2025; status: source_reviewed_not_executed; source ids: coverage-segmentnt-2025-supplement; evidence-official-da4566a88ed9fa42fdcb; source locator: Primary article Sec16–17; summary: Test chromosomes 20 and 21; validation chromosome 22; training remaining chromosomes. Test chunks with genes homologous to train/validation genes excluded using Ensembl BioMart accessed 2024-05-08; homologous distal regulatory elements not excluded. Rewire has not executed this evaluation; exact sequence manifests and checkpoint hashes remain unextracted.
entity level
suite
Related records

Suggest a correction