AIPAMPDS

Generate Antimicrobial Peptides

Generate novel antimicrobial peptides using our GPT-based models. Choose a training model, specify the peptide length and quantity, then submit to start generation.

Choose your target pathogen for optimized peptide generation
Set a maximum of 10–50 residues. Each returned peptide contains 10 residues up to this maximum.
Range: 1-1000 sequences. Large batches may take longer to process.

Supported lengths: AMP Identifier 5–298 residues; MIC Classifiers 5–50; Hemolysis Predictor 6–200. Unsupported predictions are N/A. If a required prediction is unavailable, Score is N/A and Evaluate is Incomplete. Input accepts the 20 standard amino acids and X (unknown residue); X is marked with a warning. Other non-whitespace characters are rejected. Warnings and original sequence names are included in the CSV.

📊 Scoring System Explanation

Model Scores

AMP Score: Probability of being an antimicrobial peptide (0-1)

  • Threshold: 0.998
  • Score > 0.998 → Predicted as AMP (+1 point)

E.coli Score: Probability of antimicrobial activity against E.coli (0-1)

  • Threshold: 0.9980798959732056
  • Score ≥ threshold → Active against E.coli (+1 point)
  • Only evaluated for Broad-spectrum and Gram-negative targets

S.aureus Score: Probability of antimicrobial activity against S.aureus (0-1)

  • Threshold: 0.9989857077598572
  • Score ≥ threshold → Active against S.aureus (+1 point)
  • Only evaluated for Broad-spectrum and Gram-positive targets

Hemolytic Score: Probability of non-hemolytic activity (0-1), higher is safer

  • Threshold: 0.5
  • Score ≥ 0.5 → Non-hemolytic, safe for use (+1 point)
  • Score < 0.5 → Hemolytic, potentially toxic

Overall Evaluation

Score: Cumulative score based on all evaluated criteria

Good

  • Broad-spectrum: Score = 4/4 (AMP + E.coli + S.aureus + Non-hemolytic)
  • Single target: Score = 3/3 (AMP + target activity + Non-hemolytic)
  • Recommended for further validation

Mediocre

  • Broad-spectrum: Score = 2-3/4
  • Single target: Score = 2/3
  • Acceptable but may have limitations

Bad

  • Broad-spectrum: Score = 0-1/4
  • Single target: Score = 0-1/3
  • Not recommended for further study

💡 Note:
1. The scoring system adapts based on your selected target. Broad-spectrum targets evaluate both E.coli and S.aureus (max 4 points), while specific targets only evaluate the relevant pathogen (max 3 points).
2. All submitted sequences remain in the results. Predictions outside a model’s supported length range are N/A, and incomplete evaluations are marked Incomplete.

⚠️ Important: These predictions are computational estimates. High-scoring peptides should be validated experimentally before clinical or commercial use.