rewirebio.iobenchmarks
Dataset subset

POISK co-folding binder-decoy benchmark set (22,494 complexes)

The balanced benchmark the paper folds: 11,247 patent-supported antibody-antigen pairs with no known resolved structure, each paired with a putative non-binding antibody against the same antigen.

Evaluation results

52 evaluations · 58 results. Different protocols are not a single leaderboard.

Filter evaluations

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configurationProtocol and datasetFindingEvidence and details
Configuration: AlphaFold-Multimer with full-Fv ipSAE rankingProtocol: POISK binder-decoy separation in full-Fv ipSAE, Cohen's d
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.39 Cohen's d (separation of binder and decoy full-Fv ipSAE score distributions over all 22,494 complexes)
unitless · higher

Uncertainty: Not reported by the source: Table 1 prints no spread for Cohen's d; the paper says the pairwise benchmarks run on the full set of pairs without random sampling.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

AlphaFold-Multimer on POISK binder-decoy separation in full-Fv ipSAE, Cohen's d

poisk-20261115-protocol-cohens-d-ipsae

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 1, row "AlphaFold-Multimer", column "Cohen’s d", sub-column "ipSAE"
Configuration: AlphaFold-Multimer with full-Fv ipSAE rankingProtocol: POISK binder-decoy separation in ipTM, Cohen's d
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.31 Cohen's d (separation of binder and decoy ipTM score distributions over all 22,494 complexes)
unitless · higher

Uncertainty: Not reported by the source: Table 1 prints no spread for Cohen's d; the paper says the pairwise benchmarks run on the full set of pairs without random sampling.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

AlphaFold-Multimer on POISK binder-decoy separation in ipTM, Cohen's d

poisk-20261115-protocol-cohens-d-iptm

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 1, row "AlphaFold-Multimer", column "Cohen’s d", sub-column "ipTM"
Configuration: AlphaFold-Multimer with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 10% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.846 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 10% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

AlphaFold-Multimer on POISK pairwise binder-versus-decoy accuracy at 10% coverage

poisk-20261115-protocol-pairwise-coverage-10

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "AlphaFold-Multimer", column "10%"
Configuration: AlphaFold-Multimer with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 100% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.607 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, 100% coverage of the 11,247 pairs, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

AlphaFold-Multimer on POISK pairwise binder-versus-decoy accuracy at 100% coverage

poisk-20261115-protocol-pairwise-coverage-100

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "AlphaFold-Multimer", column "100%"
Configuration: AlphaFold-Multimer with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 25% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.756 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 25% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

AlphaFold-Multimer on POISK pairwise binder-versus-decoy accuracy at 25% coverage

poisk-20261115-protocol-pairwise-coverage-25

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "AlphaFold-Multimer", column "25%"
Configuration: AlphaFold-Multimer with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 50% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.68 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 50% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 1 prints no spread for this column.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

AlphaFold-Multimer on POISK pairwise binder-versus-decoy accuracy at 50% coverage

poisk-20261115-protocol-pairwise-coverage-50

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 1, row "AlphaFold-Multimer", column "Accuracy at", sub-column "50% Pairs"
Configuration: AlphaFold-Multimer with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 50% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.679 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 50% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

AlphaFold-Multimer on POISK pairwise binder-versus-decoy accuracy at 50% coverage

poisk-20261115-protocol-pairwise-coverage-50

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "AlphaFold-Multimer", column "50%"
Configuration: AlphaFold-Multimer with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 75% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.643 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 75% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

AlphaFold-Multimer on POISK pairwise binder-versus-decoy accuracy at 75% coverage

poisk-20261115-protocol-pairwise-coverage-75

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "AlphaFold-Multimer", column "75%"
Configuration: AlphaFold-Multimer with full-Fv ipSAE rankingProtocol: POISK pairs left undecidable by identical ipSAE
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
22.9% Proportion (pairs in which binder and decoy receive numerically identical full-Fv ipSAE, over all 11,247 pairs)
percent · unknown

Uncertainty: Not applicable: A count over all 11,247 pairs, not an estimate from a sample.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

AlphaFold-Multimer on POISK pairs left undecidable by identical ipSAE

poisk-20261115-protocol-pairwise-tie-fraction

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "AlphaFold-Multimer", column "Ties"
Configuration: Boltz-2 with full-Fv ipSAE rankingProtocol: POISK binder-decoy separation in full-Fv ipSAE, Cohen's d
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.23 Cohen's d (separation of binder and decoy full-Fv ipSAE score distributions over all 22,494 complexes)
unitless · higher

Uncertainty: Not reported by the source: Table 1 prints no spread for Cohen's d; the paper says the pairwise benchmarks run on the full set of pairs without random sampling.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Boltz-2 on POISK binder-decoy separation in full-Fv ipSAE, Cohen's d

poisk-20261115-protocol-cohens-d-ipsae

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 1, row "Boltz-2", column "Cohen’s d", sub-column "ipSAE"
Configuration: Boltz-2 with full-Fv ipSAE rankingProtocol: POISK binder-decoy separation in ipTM, Cohen's d
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.2 Cohen's d (separation of binder and decoy ipTM score distributions over all 22,494 complexes)
unitless · higher

Uncertainty: Not reported by the source: Table 1 prints no spread for Cohen's d; the paper says the pairwise benchmarks run on the full set of pairs without random sampling.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Boltz-2 on POISK binder-decoy separation in ipTM, Cohen's d

poisk-20261115-protocol-cohens-d-iptm

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 1, row "Boltz-2", column "Cohen’s d", sub-column "ipTM"
Configuration: Boltz-2 with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 10% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.641 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 10% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Boltz-2 on POISK pairwise binder-versus-decoy accuracy at 10% coverage

poisk-20261115-protocol-pairwise-coverage-10

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "Boltz-2", column "10%"
Configuration: Boltz-2 with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 100% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.581 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, 100% coverage of the 11,247 pairs, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Boltz-2 on POISK pairwise binder-versus-decoy accuracy at 100% coverage

poisk-20261115-protocol-pairwise-coverage-100

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "Boltz-2", column "100%"
Configuration: Boltz-2 with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 25% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.622 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 25% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Boltz-2 on POISK pairwise binder-versus-decoy accuracy at 25% coverage

poisk-20261115-protocol-pairwise-coverage-25

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "Boltz-2", column "25%"
Configuration: Boltz-2 with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 50% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.6 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 50% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 1 prints no spread for this column.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Boltz-2 on POISK pairwise binder-versus-decoy accuracy at 50% coverage

poisk-20261115-protocol-pairwise-coverage-50

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 1, row "Boltz-2", column "Accuracy at", sub-column "50% Pairs"
Configuration: Boltz-2 with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 50% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.602 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 50% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Boltz-2 on POISK pairwise binder-versus-decoy accuracy at 50% coverage

poisk-20261115-protocol-pairwise-coverage-50

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "Boltz-2", column "50%"
Configuration: Boltz-2 with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 75% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.591 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 75% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Boltz-2 on POISK pairwise binder-versus-decoy accuracy at 75% coverage

poisk-20261115-protocol-pairwise-coverage-75

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "Boltz-2", column "75%"
Configuration: Boltz-2 with full-Fv ipSAE rankingProtocol: POISK pairs left undecidable by identical ipSAE
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
4.3% Proportion (pairs in which binder and decoy receive numerically identical full-Fv ipSAE, over all 11,247 pairs)
percent · unknown

Uncertainty: Not applicable: A count over all 11,247 pairs, not an estimate from a sample.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Boltz-2 on POISK pairs left undecidable by identical ipSAE

poisk-20261115-protocol-pairwise-tie-fraction

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "Boltz-2", column "Ties"
Configuration: Cross-model structural consensus, at least three models at DockQ 0.23Protocol: POISK pairwise binder-versus-decoy accuracy under cross-model structural consensus
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
73.1% Paired accuracy (binder identified as the pair member with more agreeing folding-model poses, at least three models agreeing at DockQ 0.23 or better, over the pairs eligible at that threshold)
percent · higher

Uncertainty: Not reported by the source: No spread is printed for the consensus accuracies.

Coverage: Not reported (Covers 54.2% of the 11,247 pairs; the eligible pair count is not printed as an integer.)

Author-reported evaluation · Source checked
Methods, coverage and source

Cross-model structural consensus (at least three models agreeing at DockQ 0.23 or better) on POISK pairwise accuracy

poisk-20261115-protocol-consensus-pairwise-accuracy

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Section 3.3.1, sentence beginning "When at least three models agree on a pose at a DockQ threshold of 0.23"
Configuration: Cross-model structural consensus, at least three models at DockQ 0.23Protocol: POISK pair coverage under cross-model structural consensus
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
54.2% Proportion (share of the 11,247 pairs eligible for classification with at least three models agreeing at DockQ 0.23 or better)
percent · unknown

Uncertainty: Not applicable: A share of the full pair set, not an estimate from a sample.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Cross-model structural consensus (at least three models agreeing at DockQ 0.23 or better) on POISK pair coverage

poisk-20261115-protocol-consensus-pair-coverage

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Section 3.3.1, sentence beginning "When at least three models agree on a pose at a DockQ threshold of 0.23"
Configuration: Cross-model structural consensus, at least three models at DockQ 0.80Protocol: POISK pairwise binder-versus-decoy accuracy under cross-model structural consensus
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
96.8% Paired accuracy (binder identified as the pair member with more agreeing folding-model poses, at least three models agreeing at DockQ 0.80 or better, over the pairs eligible at that threshold)
percent · higher

Uncertainty: Not reported by the source: No spread is printed for the consensus accuracies.

Coverage: Not reported (Covers 3.0% of the 11,247 pairs; the eligible pair count is not printed as an integer.)

Author-reported evaluation · Source checked
Methods, coverage and source

Cross-model structural consensus (at least three models agreeing at DockQ 0.80 or better) on POISK pairwise accuracy

poisk-20261115-protocol-consensus-pairwise-accuracy

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Section 3.3.1, sentence beginning "For instance, at k ≥ 3 and DockQ ≥ 0.80"
Configuration: Cross-model structural consensus, at least three models at DockQ 0.80Protocol: POISK pair coverage under cross-model structural consensus
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
3% Proportion (share of the 11,247 pairs eligible for classification with at least three models agreeing at DockQ 0.80 or better)
percent · unknown

Uncertainty: Not applicable: A share of the full pair set, not an estimate from a sample.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Cross-model structural consensus (at least three models agreeing at DockQ 0.80 or better) on POISK pair coverage

poisk-20261115-protocol-consensus-pair-coverage

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Section 3.3.1, sentence beginning "For instance, at k ≥ 3 and DockQ ≥ 0.80"
Configuration: IntelliFold v2 with full-Fv ipSAE rankingProtocol: POISK binder-decoy separation in full-Fv ipSAE, Cohen's d
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.66 Cohen's d (separation of binder and decoy full-Fv ipSAE score distributions over all 22,494 complexes)
unitless · higher

Uncertainty: Not reported by the source: Table 1 prints no spread for Cohen's d; the paper says the pairwise benchmarks run on the full set of pairs without random sampling.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

IntelliFold v2 on POISK binder-decoy separation in full-Fv ipSAE, Cohen's d

poisk-20261115-protocol-cohens-d-ipsae

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 1, row "IntelliFold v2", column "Cohen’s d", sub-column "ipSAE"
Configuration: IntelliFold v2 with full-Fv ipSAE rankingProtocol: POISK binder-decoy separation in ipTM, Cohen's d
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.6 Cohen's d (separation of binder and decoy ipTM score distributions over all 22,494 complexes)
unitless · higher

Uncertainty: Not reported by the source: Table 1 prints no spread for Cohen's d; the paper says the pairwise benchmarks run on the full set of pairs without random sampling.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

IntelliFold v2 on POISK binder-decoy separation in ipTM, Cohen's d

poisk-20261115-protocol-cohens-d-iptm

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 1, row "IntelliFold v2", column "Cohen’s d", sub-column "ipTM"
Configuration: IntelliFold v2 with full-Fv ipSAE rankingProtocol: POISK pairwise binder-versus-decoy accuracy at 10% coverage
Dataset subset: POISK co-folding binder-decoy benchmark set (22,494 complexes)
0.96 Paired accuracy (binder scored above its matched putative decoy by full-Fv ipSAE, top 10% of the 11,247 pairs ranked by within-pair maximum ipSAE, exact ties at half credit)
fraction · higher

Uncertainty: Not reported by the source: Table 7 prints no spread; the paper says the pairwise benchmarks operate on the full set of 11,247 pairs without random sampling, so error bars do not apply.

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

IntelliFold v2 on POISK pairwise binder-versus-decoy accuracy at 10% coverage

poisk-20261115-protocol-pairwise-coverage-10

Aggregation: Not reported

Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF) · Table 7, row "IntelliFold v2", column "10%"

Source checking is not independent reproduction. Release 2026-11-21-c8bea25e5abb.

Research readiness

0 of 4 readiness checks met. These checks assess whether the evidence supports a reproducible investigation; a source-checked score alone does not meet them.

Readiness checks, gaps and artifacts

Release 2026-11-21-c8bea25e5abb · Evidence verified: Not verified

Evidence incomplete

Replay metrics

Exact outcomes, predictions, identifiers and evaluator are connected.

Missing or unresolved evidence

  • No verified artifact manifest is linked to this exact record.
  • File checksums match the recorded files: not yet verified
  • Predictions are matched to the right samples: not yet verified
  • Score meaning and direction are confirmed: not yet verified
  • Metrics are recomputed from the saved predictions: not yet verified

Verified: Not verified

Evidence incomplete

Investigate discrepancies

Replay evidence includes annotations and an assessment of dependence. Unknown independence permits descriptive analysis only.

Missing or unresolved evidence

  • No verified artifact manifest is linked to this exact record.
  • File checksums match the recorded files: not yet verified
  • Predictions are matched to the right samples: not yet verified
  • Score meaning and direction are confirmed: not yet verified
  • Metrics are recomputed from the saved predictions: not yet verified
  • Sample annotations are recorded: not yet verified
  • Dependence between samples is assessed: not yet verified

Verified: Not verified

Evidence incomplete

Run locally

A pinned recipe describes the inputs, environment and resource requirements.

Missing or unresolved evidence

  • No verified artifact manifest is linked to this exact record.
  • File checksums match the recorded files: not yet verified
  • Predictions are matched to the right samples: not yet verified
  • Score meaning and direction are confirmed: not yet verified
  • A pinned run recipe exists: not yet verified
  • Compute requirements are estimated: not yet verified

Verified: Not verified

Evidence incomplete

Validate independently

Separate data and exposure records support an independent test.

Missing or unresolved evidence

  • No verified artifact manifest is linked to this exact record.
  • File checksums match the recorded files: not yet verified
  • Predictions are matched to the right samples: not yet verified
  • Score meaning and direction are confirmed: not yet verified
  • Independent validation data exist: not yet verified
  • Overlap with training data is checked: not yet verified

Verified: Not verified

Readiness describes the evidence in this release. Availability on your computer is checked separately when an investigation runs. Existing data exposure can prevent independent validation even when files are available.

Artifacts and reproduction

No verified artifact manifest is connected to this record yet. The gaps above identify what is needed before analysis can begin.

Subset and evaluation context

This record describes a particular subset or cohort used in an evaluation. Its results do not describe the full dataset.

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.

12 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-11-21-c8bea25e5abb
Property and statementOriginal source and locationReview and provenance
Limitations
3 values
  • The 11,247 negatives are putative: 7,647 are nearest-CDR matches retrieved from the Observed Antibody Space and 3,600 are germline-reverted with randomised CDR-H3. The paper states they have not been experimentally validated for lack of binding.
  • Excluding antibodies close to SAbDab reduces but does not establish absence from the models' training data; the Discussion says incomplete training-data disclosure prevents fully excluding overlap.
  • Binder labels rest on an LLM extraction from patent text, audited on a 40-record manual sample.
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.limitations

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

Positive count
11247
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.positive_count

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

Source dataset
poisk-20261115-dataset-poisk-v1-0-1
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.source_dataset

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

Source location
Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.source_locator

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

Split
Evaluation only; all 22,494 complexes are scored and none is held out
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.split

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

Subset scope
Records with a novel, human-targeting antibody and an explicit antigen sequence, with conditional (for example pH-dependent) binding excluded, antigen length between 50 and 750 residues, duplicates removed, and antibodies sharing 95% or more sequence identity with any SAbDab antibody excluded by Diamond BLASTP. Each of the 11,247 remaining binders is paired with one putative non-binder for the same antigen.
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.subset_scope

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

Test count
22494
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.test_count

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

Version
As described in bioRxiv v1 Section 2.2; folded against POISK dataset version 1.0.1
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.version

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

Description
The balanced benchmark the paper folds: 11,247 patent-supported antibody-antigen pairs with no known resolved structure, each paired with a putative non-binding antibody against the same antigen.
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: description

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

Relationship: same data as
poisk-20261115-dataset-poisk-v1-0-1
Context-only references
Tupikina et al. 2026, Can Folding Models Tell Binders from Bluffers? Evidence from POISK (bioRxiv v1 PDF)

Original source ↗

Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Version: v1, posted 26 October 2026
Retrieved: 2026-11-15T23:32:24Z

not individually reviewed

No individual claim review recorded

Audit details

Field: links:same_data_as:poisk-20261115-dataset-poisk-v1-0-1

Source artifact SHA-256: 526def150fb5ae8f9ac4533efcf16c86e547e636cdb9bfba9a574cd1e2a2f924

Hash scope: One download of 2026-11-15. The PDF container is not byte-stable: bioRxiv regenerates it on request, so artifact_sha256 is that one download's hash and no refetch reproduces it. Three downloads on 2026-11-16 gave 3389c1bf6d3654c5b2151a07fcfcefc67cf6f36e5f0facfc03c33521d8b84491 twice and d699fd680e3b0097ca80130c288e8d872773dc1e523044f4605e06fc4f89d9bc once. All three yield the same pdftotext -layout text layer, SHA-256 e75c18130ca43770d2f0ed05ae96528759666c0e4fa08f7725f5a32b21187696, which is the article's stable identity, is archived gzipped in artifacts/biorxiv-v1-layout.txt.gz under archive_sha256, is what extract/build_batch.py asserts, and is what every stored value was read from.

Format: application/pdf, 16 pages

How to read the evidence

Sources and history

Release 2026-11-21-c8bea25e5abb · Record review: source checked

1 source record and release historyDownload this release (gzip)
Technical metadata and record identifiers

Technical metadata for this record is tracked under stable ID poisk-20261115-data-cofolding-benchmark-22494.

Source locator: Section 2.2 'Co-folding benchmark data preparation'; Appendix D.1

Related records

Suggest a correction