rewire.it
Task

BEND GENE-FINDING: Gene finding

Gene finding. Scored with MCC on GENCODE. A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

16 evaluations · 16 metric rows

Overview

Gene finding. Scored with MCC on GENCODE. A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

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

Evaluation design

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

Benchmarks

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

Recorded evaluations

Each evaluation records what was tested and under which conditions.

Explore all linked results

Run instructions

No runnable recipe has been reviewed for this task. Dataset access, model requirements, licences and compute requirements must be checked against its sources before execution.

A task describes a biological question. Choose a linked protocol to obtain concrete split and scoring instructions.

Published comparisons

Explore the results reported under one evaluation protocol. Each figure keeps its source, dataset and metric together; it is not a ranking across studies. The pooled view gathers every source table that reports the same metric and names what it does not hold constant.

BEND GENE-FINDING: Gene finding

mcc (correlation) · Higher values are better for this metric.

Every method BEND reports on Gene finding, scored with MCC on GENCODE.

Evaluation protocol · GENCODE (BEND split)

  1. AUGUSTUS · Configuration · Author-reported evaluation0.80
  2. ResNet · Configuration · Author-reported evaluation0.46
  3. CNN · Method · Author-reported evaluation0.00
  4. ResNet-LM · Method · Author-reported evaluation0.36
  5. AWD-LSTM · Method · Author-reported evaluation0.05
  6. NT-H · Method · Author-reported evaluation0.41
  7. NT-MS · Method · Author-reported evaluation0.68
  8. NT-1000G · Configuration · Author-reported evaluation0.49
  9. NT-V2 · Configuration · Author-reported evaluation0.64
  10. DNABERT · Configuration · Author-reported evaluation0.20
  11. DNABERT-2 · Configuration · Author-reported evaluation0.43
  12. GENA-LM BERT · Configuration · Author-reported evaluation0.52
  13. GENA-LM BigBird · Configuration · Author-reported evaluation0.39
  14. HyenaDNA large · Configuration · Author-reported evaluation0.35
  15. HyenaDNA tiny · Configuration · Author-reported evaluation0.10
  16. GROVER · Configuration · Author-reported evaluation0.28

Source order is preserved. Plotted marks show point estimates; uncertainty, where reported, is retained in the printed values and table. Differences do not establish statistical significance.

BEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 1, row(Gene finding)
Values, uncertainty and evidence
mcc: original source values
Tested entityPrinted valueUncertaintyEvidence
AUGUSTUS · Configuration0.80 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(Expert method), column(Gene finding)
ResNet · Configuration0.46 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(ResNet), column(Gene finding)
CNN · Method0.00 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(CNN), column(Gene finding)
ResNet-LM · Method0.36 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(ResNet-LM), column(Gene finding)
AWD-LSTM · Method0.05 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(AWD-LSTM), column(Gene finding)
NT-H · Method0.41 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(NT-H), column(Gene finding)
NT-MS · Method0.68 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(NT-MS), column(Gene finding)
NT-1000G · Configuration0.49 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(NT-1000G), column(Gene finding)
NT-V2 · Configuration0.64 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(NT-V2), column(Gene finding)
DNABERT · Configuration0.20 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(DNABERT), column(Gene finding)
DNABERT-2 · Configuration0.43 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(DNABERT-2), column(Gene finding)
GENA-LM BERT · Configuration0.52 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(GENA-LM BERT), column(Gene finding)
GENA-LM BigBird · Configuration0.39 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(GENA-LM BigBird), column(Gene finding)
HyenaDNA large · Configuration0.35 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(HyenaDNA large), column(Gene finding)
HyenaDNA tiny · Configuration0.10 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(HyenaDNA tiny), column(Gene finding)
GROVER · Configuration0.28 correlationNot reportedAuthor-reported evaluation · source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(GROVER), column(Gene finding)
Scope and limitations
  • Author-reported numbers, source checked but not independently reproduced.
  • The expert entries are specialist published models, each compared on one task only.
  • The metric differs by task, taken from Table 1, so these figures cannot be averaged into one score.

Source transcription and grouping reviewed by automated source review on 2026-09-18. These experiments were not independently reproduced by rewire.

Tested entities and results

Release 2026-09-17-134cd1815de8 · 16 evaluations · 16 metric rows. Different protocols are not a single leaderboard. Where several source tables report the same metric, the published comparisons above offer a pooled view that names what it does not hold constant.

Results grouped by the exact reported evaluation
Metric and findingCoverage and uncertaintyEvidence
AUGUSTUS on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.80 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(Expert method), column(Gene finding)

Source checking is not independent reproduction.

AWD-LSTM on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.05 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(AWD-LSTM), column(Gene finding)

Source checking is not independent reproduction.

CNN on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.00 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(CNN), column(Gene finding)

Source checking is not independent reproduction.

DNABERT-2 on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.43 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(DNABERT-2), column(Gene finding)

Source checking is not independent reproduction.

DNABERT on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.20 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(DNABERT), column(Gene finding)

Source checking is not independent reproduction.

GENA-LM BERT on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.52 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(GENA-LM BERT), column(Gene finding)

Source checking is not independent reproduction.

GENA-LM BigBird on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.39 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(GENA-LM BigBird), column(Gene finding)

Source checking is not independent reproduction.

GROVER on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.28 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(GROVER), column(Gene finding)

Source checking is not independent reproduction.

HyenaDNA large on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.35 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(HyenaDNA large), column(Gene finding)

Source checking is not independent reproduction.

HyenaDNA tiny on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.10 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(HyenaDNA tiny), column(Gene finding)

Source checking is not independent reproduction.

NT-1000G on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.49 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(NT-1000G), column(Gene finding)

Source checking is not independent reproduction.

NT-H on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.41 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(NT-H), column(Gene finding)

Source checking is not independent reproduction.

NT-MS on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.68 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(NT-MS), column(Gene finding)

Source checking is not independent reproduction.

NT-V2 on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.64 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(NT-V2), column(Gene finding)

Source checking is not independent reproduction.

ResNet on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.46 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(ResNet), column(Gene finding)

Source checking is not independent reproduction.

ResNet-LM on BEND GENE-FINDING: Gene finding

A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.

Author-reported evaluation · Evaluation metadata: source checked

0.36 mcc

Unit: correlation · Direction: higher

Uncertainty: Not reported

Scored: Not reported · Eligible: Not reported

source checkedBEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks · Table 3, row(ResNet-LM), column(Gene finding)

Source checking is not 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.

1 evidence row matching the loaded filters

Claims, original sources and review scope · Release 2026-09-17-134cd1815de8
Property and statementOriginal source and locationReview and provenance
Relationship: part of

discovery-benchmark-bend

Individual claims
BEND: Benchmarking DNA Language Models on Biologically Meaningful Tasks

Original source ↗

Table 1, row(Gene finding)

Version: Primary full-text snapshot retrieved 2026-09-17; exact bytes pinned by SHA-256
Retrieved: 2026-09-17T08:06:28.183228+00:00

source checked

automated source review · 2026-09-18

Audit details

Primary-source transcription with no human sign-off and no independent reproduction.

Field: links:part_of:discovery-benchmark-bend

Claim: bend-association-gene-finding

Source artifact SHA-256: f709b6bef3120eb979c0a0e02d2582475c49410f29850ed7601ba7a700ca379d

Hash scope: Exact retrieved primary paper artifact bytes.

Inspected artifact

Sources and history

Release 2026-09-17-134cd1815de8 · Record review: source checked

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

Stable ID: bend-task-gene-finding

areas
dna-genomes
tasks
Gene finding
metric
MCC
metric direction
higher
dataset
GENCODE
protocol
A downstream head trained on frozen embeddings, except for the expert methods and the fully supervised baselines, which are trained end to end. Metric and splits are from Table 1.
source locator
Table 1, row(Gene finding)
comparison panels
id: bend-panel-gene-finding; title: BEND GENE-FINDING: Gene finding; protocol id: bend-task-gene-finding; dataset id: bend-dataset-gencode; metric: mcc; unit: correlation; direction: higher; result ids: bend-result-augustus-gene-finding-mcc; bend-result-resnet-gene-finding-mcc; bend-result-cnn-gene-finding-mcc; bend-result-resnet-lm-gene-finding-mcc; bend-result-awd-lstm-gene-finding-mcc; bend-result-nt-h-gene-finding-mcc; bend-result-nt-ms-gene-finding-mcc; bend-result-nt-1000g-gene-finding-mcc; bend-result-nt-v2-gene-finding-mcc; bend-result-dnabert-gene-finding-mcc; bend-result-dnabert-2-gene-finding-mcc; bend-result-gena-lm-bert-gene-finding-mcc; bend-result-gena-lm-bigbird-gene-finding-mcc; bend-result-hyenadna-large-gene-finding-mcc; bend-result-hyenadna-tiny-gene-finding-mcc; bend-result-grover-gene-finding-mcc; source ids: evidence-expansion-bend-final-f709b6be; source locator: Table 1, row(Gene finding); context: Every method BEND reports on Gene finding, scored with MCC on GENCODE.; caveats: Author-reported numbers, source checked but not independently reproduced.; The expert entries are specialist published models, each compared on one task only.; The metric differs by task, taken from Table 1, so these figures cannot be averaged into one score.; review: method: automated_source_review; date: 2026-09-18
Related records

Suggest a correction