Evidence claim

ProteinBench against ProteinGym on ESM-IF1 and ProteinMPNN: different tasks

Discrepancy review of ESM-IF1 · ProteinBench IF-DE-NOVO-BACKBONES-BASED-SEQUENCE-DESIGN-LENGTH-100-PLDDT · pLDDT

explained

Review of discrepancy-1f1fd7fb7ee75052, revision 2: explained. Discrepancy rules version 4 gave discrepancy-d734930095432234 a new ID (two-study-conflict); the state and resolution are carried forward unchanged.

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.

31 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-11-02-a7f03c9e55a5
Property and statement	Original source and location	Review and provenance
Date 2026-10-24 Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `attributes.date` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact
Discrepancy id discrepancy-1f1fd7fb7ee75052 Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `attributes.discrepancy_id` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact
Claimed field discrepancy_review Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `attributes.field` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact
Investigation url https://github.com/rewire-bio/rewire-benchmark-data/issues/201 Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `attributes.investigation_url` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact
Previous ids 1 values discrepancy-d734930095432234 Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `attributes.previous_ids` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact
Rationale ProteinBench measures sequence design on de novo backbones of length 100 to 500, scored by pLDDT and scTM after refolding, the task ProteinMPNN was built for. Every consensus comparison is ProteinGym zero-shot variant effect prediction (DMS Spearman, clinical AUC), where ESM-IF1 scores 0.422 against ProteinMPNN's 0.257. Design quality and variant scoring are different endpoints; both sets of results are kept as reported. The high severity comes from the strong-consensus step cancelling the cross-task step (issue 201). Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `attributes.rationale` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact
Revision 2 Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `attributes.revision` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact
Source location The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `attributes.source_locator` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact
Claimed value explained Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `attributes.value` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact
Description Review of discrepancy-1f1fd7fb7ee75052, revision 2: explained. Discrepancy rules version 4 gave discrepancy-d734930095432234 a new ID (two-study-conflict); the state and resolution are carried forward unchanged. Context-only references	proteinbench primary benchmark evidence Original source ↗ The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6) Version: 2409.06744v1 Retrieved: 2026-09-16T21:07:13.231727+00:00	not individually reviewed No individual claim review recorded Audit details Field: `description` Source artifact SHA-256: `4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact

How to read the evidence

Sources and history

Release 2026-11-02-a7f03c9e55a5 · Record review: source checked

1 source record and release history

proteinbench primary benchmark evidence · Original source · 2409.06744v1

Download this release (gzip)

Technical metadata and extraction receipts

Stable ID: discrepancy-review-20261023-proteinbench-esmif1-proteinmpnn-r2

areas: proteins-complexes
contexts: research
field: discrepancy_review
discrepancy id: discrepancy-1f1fd7fb7ee75052
revision: 2
value: explained
date: 2026-10-24
rationale: ProteinBench measures sequence design on de novo backbones of length 100 to 500, scored by pLDDT and scTM after refolding, the task ProteinMPNN was built for. Every consensus comparison is ProteinGym zero-shot variant effect prediction (DMS Spearman, clinical AUC), where ESM-IF1 scores 0.422 against ProteinMPNN's 0.257. Design quality and variant scoring are different endpoints; both sets of results are kept as reported. The high severity comes from the strong-consensus step cancelling the cross-task step (issue 201).
investigation url: https://github.com/rewire-bio/rewire-benchmark-data/issues/201
source locator: The outlier and consensus results of discrepancy-d734930095432234 as published at https://benchmarks.rewirebio.io/contested/discrepancy-d734930095432234/ (release 2026-10-23-968c8720eea6)
review: method: automated-source-review; ai-assisted-source-review; reviewer: claude; reviewer note: Recorded by a Claude agent; no human review claimed.; date: 2026-10-24; method note: Read every outlier and consensus result of the finding, with its protocol, metric, configuration and source, from the published contested-result page, and compared the tasks, settings and configurations on each side. The original papers were not re-opened.; note: Carried forward by scripts/kg/migrate-discrepancy-ids.py when rules version 4 re-keyed the finding; the evidence and decision are those of discrepancy-review-20261023-proteinbench-esmif1-proteinmpnn.
previous ids: discrepancy-d734930095432234

Related records

Suggest a correction