※ GPS-SUMO 2.0 Online Service (Penalized Logistic Regression) 
For prediction based on transformer neural network, please click here.
For comprehensive prediction with all models and annotations of secondary structure and surface accessibility, please click here.
For species-specific prediction, please click here.
For prediction according to protein identiter, please click here.
Please enter sequence(s) in FASTA format
Or, upload file (<70K)
All the spaces,
line breaks will be automatically removed. You could input one
primary sequence or
multiple proteins' sequences
in FASTA format. And please don't input any special characters.
After GPS-SUMO 2.0 predictor model was well-trained, we performed an evaluation on this model. From the evaluation, three thresholds with High, Medium and Low stringency were chosen for GPS-SUMO 2.0. The performance under these three thresholds was presented as follow:
| Threshold | Sumoylation | SUMO interaction | ||||||||
| Ac | Sn | Sp | MCC | Pr | Ac | Sn | Sp | MCC | Pr | |
| High | 76.16% | 57.24% | 95.08% | 0.5652 | 92.08% | 85.83% | 67.50% | 95.00% | 0.6731 | 87.10% |
| Medium | 78.75% | 67.34% | 90.16% | 0.5906 | 87.25% | 92.50% | 97.50% | 90.00% | 0.8450 | 82.98% |
| Low | 80.38% | 75.76% | 85.00% | 0.6102 | 80.38% | 90.00% | 98.75% | 85.62% | 0.8046 | 77.45% |