# Contact Prediction on CAMEO-PTA25 L ≥ 75, N=29 2022-09 cutoff

Source: [Contact Prediction on CAMEO-PTA25 L ≥ 75, N=29 2022-09 cutoff](<https://www.sota2.com/research/sota/contact-prediction-on-cameo-pta25-l-75-n-29-2022-09-cutoff>)

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

Data updated: 2026-06-23T09:47:37.424Z

Evaluation of Contact Prediction performance on the CAMEO-PTA25 dataset (L ≥ 75, N=29, 2022-09 cutoff).

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 | Contact Prediction |
| Dataset (including any stated split) | CAMEO-PTA25 L ≥ 75, N=29 2022-09 cutoff |
| Metric | Precision @ L/2 (long) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 14 of 14 |
| 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

- [Precision @ L/2 (long)](<https://www.sota2.com/research/sota/contact-prediction-on-cameo-pta25-l-75-n-29-2022-09-cutoff.md?metric=Precision%20%40%20L%2F2%20(long)>): Higher is better; unit: PERCENT.

- [Improvement (fusion - CJ)](<https://www.sota2.com/research/sota/contact-prediction-on-cameo-pta25-l-75-n-29-2022-09-cutoff.md?metric=Improvement%20(fusion%20-%20CJ)>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Precision @ L/2 (long) | Result date | Report paper |
| --- | --- | --- | --- |
| top-K attention fusion | 46.3 | Not recorded | 6a3a1326206542251d7c70e3 |
| top-K attention fusion | 45.9 | Not recorded | 6a3a1326206542251d7c70e3 |
| top-K attention fusion | 41.8 | Not recorded | 6a3a1326206542251d7c70e3 |
| top-1 head | 39.6 | Not recorded | 6a3a1326206542251d7c70e3 |
| Co-evolutionary Jacobian (CJ) | 36.9 | Not recorded | 6a3a1326206542251d7c70e3 |
| Co-evolutionary Jacobian (CJ) | 36.9 | Not recorded | 6a3a1326206542251d7c70e3 |
| Co-evolutionary Jacobian (CJ) | 35.9 | Not recorded | 6a3a1326206542251d7c70e3 |
| top-1 head | 35.4 | Not recorded | 6a3a1326206542251d7c70e3 |
| top-K attention fusion | 34.8 | Not recorded | 6a3a1326206542251d7c70e3 |
| top-1 head | 34.7 | Not recorded | 6a3a1326206542251d7c70e3 |
| top-1 head | 34 | Not recorded | 6a3a1326206542251d7c70e3 |
| Co-evolutionary Jacobian (CJ) | 28.9 | Not recorded | 6a3a1326206542251d7c70e3 |
| top-K attention fusion | 26.2 | Not recorded | 6a3a1326206542251d7c70e3 |
| top-1 head | 25.5 | Not recorded | 6a3a1326206542251d7c70e3 |

### top-K attention fusion

Precision @ L/2 (long): 46.3. Result ID: 6a3a5628fbc373592c361aa9.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ESM-2-3B","K":"10","fusion\_mode":"naive-mean"}. This record does not assert independent reproduction.

### top-K attention fusion

Precision @ L/2 (long): 45.9. Result ID: 6a3a5628fbc373592c361aa6.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ESM-2-650M","K":"10","fusion\_mode":"naive-mean"}. This record does not assert independent reproduction.

### top-K attention fusion

Precision @ L/2 (long): 41.8. Result ID: 6a3a5628fbc373592c361aac.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ProtT5-XL","K":"10","fusion\_mode":"naive-mean"}. This record does not assert independent reproduction.

### top-1 head

Precision @ L/2 (long): 39.6. Result ID: 6a3a5628fbc373592c361aa5.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ESM-2-650M","K":"10"}. This record does not assert independent reproduction.

### Co-evolutionary Jacobian (CJ)

Precision @ L/2 (long): 36.9. Result ID: 6a3a5628fbc373592c361aa7.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ESM-2-650M","K":"10","readout\_type":"logit"}. This record does not assert independent reproduction.

### Co-evolutionary Jacobian (CJ)

Precision @ L/2 (long): 36.9. Result ID: 6a3a5628fbc373592c361aaa.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ESM-2-3B","K":"10","readout\_type":"logit"}. This record does not assert independent reproduction.

### Co-evolutionary Jacobian (CJ)

Precision @ L/2 (long): 35.9. Result ID: 6a3a5628fbc373592c361aad.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ProtT5-XL","K":"10","readout\_type":"repr"}. This record does not assert independent reproduction.

### top-1 head

Precision @ L/2 (long): 35.4. Result ID: 6a3a5628fbc373592c361aab.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ProtT5-XL","K":"10"}. This record does not assert independent reproduction.

### top-K attention fusion

Precision @ L/2 (long): 34.8. Result ID: 6a3a5628fbc373592c361aaf.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ESM-1b","K":"10","fusion\_mode":"naive-mean"}. This record does not assert independent reproduction.

### top-1 head

Precision @ L/2 (long): 34.7. Result ID: 6a3a5628fbc373592c361aa8.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ESM-2-3B","K":"10"}. This record does not assert independent reproduction.

### top-1 head

Precision @ L/2 (long): 34. Result ID: 6a3a5628fbc373592c361aae.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ESM-1b","K":"10"}. This record does not assert independent reproduction.

### Co-evolutionary Jacobian (CJ)

Precision @ L/2 (long): 28.9. Result ID: 6a3a5628fbc373592c361ab0.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"ESM-1b","K":"10","readout\_type":"logit"}. This record does not assert independent reproduction.

### top-K attention fusion

Precision @ L/2 (long): 26.2. Result ID: 6a3a5628fbc373592c361ab2.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"AMPLIFY-350M†","K":"10","leakage\_cleanliness":"partial","fusion\_mode":"naive-mean"}. This record does not assert independent reproduction.

### top-1 head

Precision @ L/2 (long): 25.5. Result ID: 6a3a5628fbc373592c361ab1.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1326206542251d7c70e3.md>)

Recorded evaluation configuration: {"Architecture":"AMPLIFY-350M†","K":"10","leakage\_cleanliness":"partial"}. This record does not assert independent reproduction.

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