# Protein Sequence Generation on Protein Safety Evaluation Harmful Proteins

Source: [Protein Sequence Generation on Protein Safety Evaluation Harmful Proteins](<https://www.sota2.com/research/sota/protein-sequence-generation-on-protein-safety-evaluation-harmful-proteins>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:36:45.068Z

Evaluation of generative models for producing safe protein sequences, utilizing metrics derived from sequence alignment (BLAST, MMseq2) and toxicity/structure prediction tools (Pfam, ToxinPred3).

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Protein Sequence Generation |
| Dataset (including any stated split) | Protein Safety Evaluation Harmful Proteins |
| Metric | BLAST Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [BLAST Score](<https://www.sota2.com/research/sota/protein-sequence-generation-on-protein-safety-evaluation-harmful-proteins.md?metric=BLAST%20Score>): Higher is better; unit: SCALAR.

- [MMseq2 Score](<https://www.sota2.com/research/sota/protein-sequence-generation-on-protein-safety-evaluation-harmful-proteins.md?metric=MMseq2%20Score>): Higher is better; unit: SCALAR.

- [Pfam Domain Score](<https://www.sota2.com/research/sota/protein-sequence-generation-on-protein-safety-evaluation-harmful-proteins.md?metric=Pfam%20Domain%20Score>): Higher is better; unit: SCALAR.

- [Pfam Evaluation Score](<https://www.sota2.com/research/sota/protein-sequence-generation-on-protein-safety-evaluation-harmful-proteins.md?metric=Pfam%20Evaluation%20Score>): Higher is better; unit: SCALAR.

- [Toxicity Risk Score](<https://www.sota2.com/research/sota/protein-sequence-generation-on-protein-safety-evaluation-harmful-proteins.md?metric=Toxicity%20Risk%20Score>): Lower is better; unit: SCALAR.

## Recorded results

| Method | BLAST Score | Result date | Report paper |
| --- | --- | --- | --- |
| InstructProtein | 0.41 | Not recorded | 68fa7a101e209131cea80e2c |
| ProtGPT2 | 0.269 | Not recorded | 68fa7a101e209131cea80e2c |
| Progen2 | 0.155 | Not recorded | 68fa7a101e209131cea80e2c |
| ProtGPT2+KPO | 0.138 | Not recorded | 68fa7a101e209131cea80e2c |
| Progen2+KPO | 0.128 | Not recorded | 68fa7a101e209131cea80e2c |
| InstructProtein+KPO | 0.086 | Not recorded | 68fa7a101e209131cea80e2c |

### InstructProtein

BLAST Score: 0.41. Result ID: 698b6c027a967eddb76e739f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa5f511e209131ce9e8f59.md>)

Recorded evaluation configuration: {"Base Model":"InstructProtein"}. This record does not assert independent reproduction.

### ProtGPT2

BLAST Score: 0.269. Result ID: 698b6c027a967eddb76e739b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c.md>)

Recorded evaluation configuration: {"Base Model":"ProtGPT2"}. This record does not assert independent reproduction.

### Progen2

BLAST Score: 0.155. Result ID: 698b6c027a967eddb76e739d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa5f2a1e209131ce9e7f7a.md>)

Recorded evaluation configuration: {"Base Model":"Progen2"}. This record does not assert independent reproduction.

### ProtGPT2+KPO

BLAST Score: 0.138. Result ID: 698b6c027a967eddb76e739c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c.md>)

[Implementation](<https://github.com/HICAI-ZJU/KPO>)

Recorded evaluation configuration: {"Base Model":"ProtGPT2"}. This record does not assert independent reproduction.

### Progen2+KPO

BLAST Score: 0.128. Result ID: 698b6c027a967eddb76e739e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c.md>)

[Implementation](<https://github.com/HICAI-ZJU/KPO>)

Recorded evaluation configuration: {"Base Model":"Progen2"}. This record does not assert independent reproduction.

### InstructProtein+KPO

BLAST Score: 0.086. Result ID: 698b6c027a967eddb76e73a0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c.md>)

[Implementation](<https://github.com/HICAI-ZJU/KPO>)

Recorded evaluation configuration: {"Base Model":"InstructProtein"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.