PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
ProteinGym zero-shot variant effect prediction. Scored with Spearman on ProteinGym. Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4.
Overview
ProteinGym zero-shot variant effect prediction. Scored with Spearman on ProteinGym. Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4.
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.
- DPLM on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- ESM-2 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- ESM-C on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- ESM3 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- PGLM on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- ProtGPT2 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- ProTrek on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- SaProt on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- VenusPLM on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
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.
PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
spearman (correlation) · Higher values are better for this metric.
Every method PFMBench reports on ProteinGym zero-shot variant effect prediction, scored with Spearman on ProteinGym.
- SaProt · Configuration · Author-reported evaluation0.45094
- VenusPLM · Configuration · Author-reported evaluation0.43952
- ESM-2 · Configuration · Author-reported evaluation0.43904
- ESM-C · Configuration · Author-reported evaluation0.43422
- DPLM · Configuration · Author-reported evaluation0.42922
- ESM3 · Configuration · Author-reported evaluation0.41401
- PGLM · Configuration · Author-reported evaluation0.39750
- ProTrek · Configuration · Author-reported evaluation0.35919
- ProtGPT2 · Configuration · Author-reported evaluation0.18962
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.
pfmbench primary benchmark evidence · Table 1, row(ProteinGym)Values, uncertainty and evidence
| Tested entity | Printed value | Uncertainty | Evidence |
|---|---|---|---|
| SaProt · Configuration | 0.45094 correlation | Not reported | Author-reported evaluation · source checkedpfmbench primary benchmark evidence · Table 4, row(SaProt [ 55 ]), column(ProteinGym) |
| VenusPLM · Configuration | 0.43952 correlation | Not reported | Author-reported evaluation · source checkedpfmbench primary benchmark evidence · Table 4, row(VenusPLM [ 59 ]), column(ProteinGym) |
| ESM-2 · Configuration | 0.43904 correlation | Not reported | Author-reported evaluation · source checkedpfmbench primary benchmark evidence · Table 4, row(ESM-2 [ 35 ]), column(ProteinGym) |
| ESM-C · Configuration | 0.43422 correlation | Not reported | Author-reported evaluation · source checkedpfmbench primary benchmark evidence · Table 4, row(ESM-C), column(ProteinGym) |
| DPLM · Configuration | 0.42922 correlation | Not reported | Author-reported evaluation · source checkedpfmbench primary benchmark evidence · Table 4, row(DPLM [ 68 ]), column(ProteinGym) |
| ESM3 · Configuration | 0.41401 correlation | Not reported | Author-reported evaluation · source checkedpfmbench primary benchmark evidence · Table 4, row(ESM3 [ 18 ]), column(ProteinGym) |
| PGLM · Configuration | 0.39750 correlation | Not reported | Author-reported evaluation · source checkedpfmbench primary benchmark evidence · Table 4, row(PGLM [ 5 ]), column(ProteinGym) |
| ProTrek · Configuration | 0.35919 correlation | Not reported | Author-reported evaluation · source checkedpfmbench primary benchmark evidence · Table 4, row(ProTrek [ 56 ]), column(ProteinGym) |
| ProtGPT2 · Configuration | 0.18962 correlation | Not reported | Author-reported evaluation · source checkedpfmbench primary benchmark evidence · Table 4, row(ProtGPT2 [ 12 ]), column(ProteinGym) |
Scope and limitations
- Author-reported numbers, source checked but not independently reproduced.
- The metric differs by task, taken from Table 1, so these figures cannot be averaged into one score.
- Table 3 scores come from adapter fine-tuning; the ProteinGym figure is zero-shot and is not comparable to them.
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 · 9 evaluations · 9 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.
| Metric and finding | Coverage and uncertainty | Evidence |
|---|---|---|
| DPLM on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: DPLMTask: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.42922 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(DPLM [ 68 ]), column(ProteinGym) Source checking is not independent reproduction. |
| ESM-2 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: ESM-2Task: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.43904 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(ESM-2 [ 35 ]), column(ProteinGym) Source checking is not independent reproduction. |
| ESM-C on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: ESM-CTask: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.43422 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(ESM-C), column(ProteinGym) Source checking is not independent reproduction. |
| ESM3 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: ESM3Task: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.41401 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(ESM3 [ 18 ]), column(ProteinGym) Source checking is not independent reproduction. |
| PGLM on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: PGLMTask: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.39750 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(PGLM [ 5 ]), column(ProteinGym) Source checking is not independent reproduction. |
| ProtGPT2 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: ProtGPT2Task: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.18962 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(ProtGPT2 [ 12 ]), column(ProteinGym) Source checking is not independent reproduction. |
| ProTrek on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: ProTrekTask: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.35919 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(ProTrek [ 56 ]), column(ProteinGym) Source checking is not independent reproduction. |
| SaProt on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: SaProtTask: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.45094 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(SaProt [ 55 ]), column(ProteinGym) Source checking is not independent reproduction. |
| VenusPLM on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: VenusPLMTask: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.43952 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(VenusPLM [ 59 ]), column(ProteinGym) 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
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Relationship: part of discovery-benchmark-pfmbench Individual claims | pfmbench primary benchmark evidence Table 1, row(ProteinGym) Version: 2506.14796v1 | source checked automated source review · 2026-09-18 Audit detailsPrimary-source transcription with no human sign-off and no independent reproduction. Field: Claim: pfmbench-association-proteingym-zs Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
Sources and history
Release 2026-09-17-134cd1815de8 · Record review: source checked
1 source records and release history
- pfmbench primary benchmark evidence · Original source · 2506.14796v1
Technical metadata and extraction receipts
Stable ID: pfmbench-task-proteingym-zs
- areas
- proteins-complexes
- tasks
- ProteinGym zero-shot variant effect prediction
- metric
- Spearman
- metric direction
- higher
- dataset
- ProteinGym
- protocol
- Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4.
- source locator
- Table 1, row(ProteinGym)
- comparison panels
- id: pfmbench-panel-proteingym-zs; title: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction; protocol id: pfmbench-task-proteingym-zs; dataset id: pfmbench-dataset-proteingym; metric: spearman; unit: correlation; direction: higher; result ids: pfmbench-result-saprot-proteingym-zs-spearman; pfmbench-result-venusplm-proteingym-zs-spearman; pfmbench-result-esm-2-proteingym-zs-spearman; pfmbench-result-esm-c-proteingym-zs-spearman; pfmbench-result-dplm-proteingym-zs-spearman; pfmbench-result-esm3-proteingym-zs-spearman; pfmbench-result-pglm-proteingym-zs-spearman; pfmbench-result-protrek-proteingym-zs-spearman; pfmbench-result-protgpt2-proteingym-zs-spearman; source ids: expansion-p3-pfmbench; source locator: Table 1, row(ProteinGym); context: Every method PFMBench reports on ProteinGym zero-shot variant effect prediction, scored with Spearman on ProteinGym.; caveats: Author-reported numbers, source checked but not independently reproduced.; The metric differs by task, taken from Table 1, so these figures cannot be averaged into one score.; Table 3 scores come from adapter fine-tuning; the ProteinGym figure is zero-shot and is not comparable to them.; review: method: automated_source_review; date: 2026-09-18
Related records
- part of: PFMBench
- subject: PFMBench PROTEINGYM-ZS: part of discovery-benchmark-pfmbench
- benchmark: DPLM on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- benchmark: ESM-2 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- benchmark: ESM-C on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- benchmark: ESM3 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- benchmark: PGLM on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- benchmark: ProtGPT2 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- benchmark: ProTrek on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- benchmark: SaProt on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- benchmark: VenusPLM on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction