BacBench strain clustering, average silhouette width
BacBench strain clustering, average silhouette width as measured in the Bacformer preprint benchmark.
Overview
An explanatory profile has not yet been reviewed for this record.
Consult the linked sources for architecture or protocol details. Missing evidence is not evidence of a missing capability.
6 recorded evaluations, 6 metric rows. A comparison chart has not yet been validated for these results. The table retains the individual findings and their sources.
Results
Results are available, but no reviewed comparison panel is linked in this release.
All evaluations
6 evaluations · 6 results. Different protocols are not a single leaderboard.
Protocol: BacBench strain clustering, average silhouette width · Dataset: BacBench strain clustering dataset of 6,071 MGnify MAGs
Sorted by Silhouette score (average silhouette width of the Leiden clustering of mean-pooled genome embeddings, at the resolution and neighbour count that maximise the mean of the three reported metrics) (higher is better). The best value in each column is highlighted. Decimals are rounded for display; each value links to the printed value and its source.
| Tested configuration | Silhouette score (average silhouette width of the Leiden clustering of mean-pooled genome embeddings, at the resolution and neighbour count that maximise the mean of the three reported metrics) |
|---|---|
| Bacformer Large genome embeddings for strain clustering (checkpoint not resolved)Evaluation | 0.460 ± 0.031 (best in column) |
| gLM2 (frozen embeddings, no trained head)Evaluation | 0.381 ± 0.012 |
| ProtBERT (frozen embeddings, no trained head)Evaluation | 0.326 ± 0.037 |
| ESM-C (frozen embeddings, no trained head)Evaluation | 0.277 ± 0.031 |
| Bacformer genome embeddings for strain clustering (checkpoint not resolved)Evaluation | 0.250 ± 0.022 |
| ESM-2 (frozen embeddings, no trained head)Evaluation | 0.191 ± 0.022 |
All 6 result rows with coverage, uncertainty and sources
Filter evaluations
Applied filters: All linked evaluations
| Tested configuration | Protocol and dataset | Finding | Evidence and details |
|---|---|---|---|
| Configuration: Bacformer genome embeddings for strain clustering (checkpoint not resolved) | Protocol: BacBench strain clustering, average silhouette width Dataset: BacBench strain clustering dataset of 6,071 MGnify MAGs | 0.250 ± 0.022 Silhouette score (average silhouette width of the Leiden clustering of mean-pooled genome embeddings, at the resolution and neighbour count that maximise the mean of the three reported metrics) unitless · higher Uncertainty: SD 0.022 Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceBacformer on BacBench strain clustering, average silhouette width Final-layer hidden states are mean-pooled into one vector per genome and clustered with Leiden over a grid of resolutions 0.1, 0.25, 0.5 and 1.0 and neighbour counts 5, 10 and 15; the combination that maximises the mean of the adjusted Rand index, normalised mutual information and average silhouette width at the species level is retained. Clustering is run with Scanpy. Aggregation: Not reported Bacformer preprint, bioRxiv 2025.07.20.665723 version 3 · Table S12, row 'Bacformer', column 'ASW' |
| Configuration: Bacformer Large genome embeddings for strain clustering (checkpoint not resolved) | Protocol: BacBench strain clustering, average silhouette width Dataset: BacBench strain clustering dataset of 6,071 MGnify MAGs | 0.460 ± 0.031 Silhouette score (average silhouette width of the Leiden clustering of mean-pooled genome embeddings, at the resolution and neighbour count that maximise the mean of the three reported metrics) unitless · higher Uncertainty: SD 0.031 Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceBacformer Large on BacBench strain clustering, average silhouette width Final-layer hidden states are mean-pooled into one vector per genome and clustered with Leiden over a grid of resolutions 0.1, 0.25, 0.5 and 1.0 and neighbour counts 5, 10 and 15; the combination that maximises the mean of the adjusted Rand index, normalised mutual information and average silhouette width at the species level is retained. Clustering is run with Scanpy. Aggregation: Not reported Bacformer preprint, bioRxiv 2025.07.20.665723 version 3 · Table S12, row 'Bacformer Large', column 'ASW' |
| Configuration: ESM-2 (frozen embeddings, no trained head) | Protocol: BacBench strain clustering, average silhouette width Dataset: BacBench strain clustering dataset of 6,071 MGnify MAGs | 0.191 ± 0.022 Silhouette score (average silhouette width of the Leiden clustering of mean-pooled genome embeddings, at the resolution and neighbour count that maximise the mean of the three reported metrics) unitless · higher Uncertainty: SD 0.022 Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM-2 on BacBench strain clustering, average silhouette width Final-layer hidden states are mean-pooled into one vector per genome and clustered with Leiden over a grid of resolutions 0.1, 0.25, 0.5 and 1.0 and neighbour counts 5, 10 and 15; the combination that maximises the mean of the adjusted Rand index, normalised mutual information and average silhouette width at the species level is retained. Clustering is run with Scanpy. Aggregation: Not reported Bacformer preprint, bioRxiv 2025.07.20.665723 version 3 · Table S12, row 'ESM-2', column 'ASW' |
| Configuration: ESM-C (frozen embeddings, no trained head) | Protocol: BacBench strain clustering, average silhouette width Dataset: BacBench strain clustering dataset of 6,071 MGnify MAGs | 0.277 ± 0.031 Silhouette score (average silhouette width of the Leiden clustering of mean-pooled genome embeddings, at the resolution and neighbour count that maximise the mean of the three reported metrics) unitless · higher Uncertainty: SD 0.031 Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM-C on BacBench strain clustering, average silhouette width Final-layer hidden states are mean-pooled into one vector per genome and clustered with Leiden over a grid of resolutions 0.1, 0.25, 0.5 and 1.0 and neighbour counts 5, 10 and 15; the combination that maximises the mean of the adjusted Rand index, normalised mutual information and average silhouette width at the species level is retained. Clustering is run with Scanpy. Aggregation: Not reported Bacformer preprint, bioRxiv 2025.07.20.665723 version 3 · Table S12, row 'ESM-C', column 'ASW' |
| Configuration: gLM2 (frozen embeddings, no trained head) | Protocol: BacBench strain clustering, average silhouette width Dataset: BacBench strain clustering dataset of 6,071 MGnify MAGs | 0.381 ± 0.012 Silhouette score (average silhouette width of the Leiden clustering of mean-pooled genome embeddings, at the resolution and neighbour count that maximise the mean of the three reported metrics) unitless · higher Uncertainty: SD 0.012 Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourcegLM2 on BacBench strain clustering, average silhouette width Final-layer hidden states are mean-pooled into one vector per genome and clustered with Leiden over a grid of resolutions 0.1, 0.25, 0.5 and 1.0 and neighbour counts 5, 10 and 15; the combination that maximises the mean of the adjusted Rand index, normalised mutual information and average silhouette width at the species level is retained. Clustering is run with Scanpy. Aggregation: Not reported Bacformer preprint, bioRxiv 2025.07.20.665723 version 3 · Table S12, row 'gLM2', column 'ASW' |
| Configuration: ProtBERT (frozen embeddings, no trained head) | Protocol: BacBench strain clustering, average silhouette width Dataset: BacBench strain clustering dataset of 6,071 MGnify MAGs | 0.326 ± 0.037 Silhouette score (average silhouette width of the Leiden clustering of mean-pooled genome embeddings, at the resolution and neighbour count that maximise the mean of the three reported metrics) unitless · higher Uncertainty: SD 0.037 Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceProtBERT on BacBench strain clustering, average silhouette width Final-layer hidden states are mean-pooled into one vector per genome and clustered with Leiden over a grid of resolutions 0.1, 0.25, 0.5 and 1.0 and neighbour counts 5, 10 and 15; the combination that maximises the mean of the adjusted Rand index, normalised mutual information and average silhouette width at the species level is retained. Clustering is run with Scanpy. Aggregation: Not reported Bacformer preprint, bioRxiv 2025.07.20.665723 version 3 · Table S12, row 'ProtBERT', column 'ASW' |
Source checking is not independent reproduction. Release 2026-11-21-adc15377fab2.
Methods and evaluation design
Procedure, tasks and evaluated configurations
Recorded evaluations
Each evaluation records what was tested and under which conditions.
- Bacformer on BacBench strain clustering, average silhouette width
- Bacformer Large on BacBench strain clustering, average silhouette width
- ESM-2 on BacBench strain clustering, average silhouette width
- ESM-C on BacBench strain clustering, average silhouette width
- gLM2 on BacBench strain clustering, average silhouette width
- ProtBERT on BacBench strain clustering, average silhouette width
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 2 active baseline roles have published Rewire measurements in this release. Measurements on a selected protocol do not establish coverage of an entire suite.
No execution recipe linked to this protocol. Recipe availability does not establish a completed evaluation.
- Author-reported evaluations
- 6
Literature evidence is not a Rewire measurement. Executed but unpublished runs and private review status are not included.
Null control
Proposed control: requires review
Select a task-valid null control after reviewing inputs and metric
Protocol-specific applicability, permitted inputs, access, split, evaluator and execution requirements need review before implementation or execution.
This is a suggested selection rule, not a validated method or a measured score.
Conventional reference
Proposed control: requires review
Select an upstream conventional reference after reviewing the full protocol
Protocol-specific applicability, permitted inputs, access, split, evaluator and execution requirements need review before implementation or execution.
This is a suggested selection rule, not a validated method or a measured score.
Protocol coverage CSV (gzip) · Model evaluation matrix (gzip) · Source table (gzip) · Release and checksums (gzip)
Coverage is derived from release 2026-11-21-adc15377fab2. Source citations describe the original records; they do not validate an unreviewed baseline proposal. No results have been generated by this audit.
Run instructions
No runnable recipe has been reviewed for this protocol. Dataset access, model requirements, licences and compute requirements must be checked against its sources before execution.
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
| Property and statement | Original source and location | Review and provenance |
|---|
No evidence rows match these filters. Choose another scope or clear the search.
Sources and history
Release 2026-11-21-adc15377fab2 · Record review: source checked
1 source record and release history
- Bacformer preprint, bioRxiv 2025.07.20.665723 version 3 · Original source · Version 3, posted 17 August 2026 according to the page banner; the bioRxiv API lists version 3 with date 2026-08-19
Technical metadata and record identifiers
Technical metadata for this record is tracked under stable ID bacformer-20261115-protocol-strain-clustering-asw.
Source locator: Table S12, column 'ASW'
Related records
- part of: BacBench
- evaluates task: BacBench Strain clustering
- uses data: BacBench strain clustering dataset of 6,071 MGnify MAGs
- assessment: Bacformer on BacBench strain clustering, average silhouette width
- assessment: Bacformer Large on BacBench strain clustering, average silhouette width
- assessment: ESM-2 on BacBench strain clustering, average silhouette width
- assessment: ESM-C on BacBench strain clustering, average silhouette width
- assessment: gLM2 on BacBench strain clustering, average silhouette width
- assessment: ProtBERT on BacBench strain clustering, average silhouette width