Peptide Agent Docs
Everything you need to know about using the AI-powered peptide discovery platform.
Overview
Peptide Agent is an AI-powered tool for early-stage peptide drug discovery. You describe a health condition or biological target in plain English, and the system generates a synthetic peptide candidate with predicted properties including binding mechanism, stability, and toxicity profile.
The platform is designed for researchers, biotech founders, and curious minds who want a fast, explainable starting point for peptide hypothesis generation. All results are AI-predicted and require laboratory validation before any clinical use.
Quick Start
- Enter a condition or target
Type a disease, condition, or biological target into the search field. Examples:
acne treatment,anti-inflammatory peptide,wound healing,GLP-1 receptor agonist. - Click Discover Peptide
The AI analyzes your query, searches known peptide data, and generates a candidate sequence with supporting analysis. This typically takes 3–6 seconds.
- Review the output
You'll receive a peptide sequence, confidence score, mechanism tag, stability and toxicity predictions, and a similarity table comparing against known peptides from UniProt.
- Save and share
Click Save Result to store the result and generate a permanent shareable link (e.g.
app.peptideagent.xyz/result/[id]).
Rate Limits
Each IP address is limited to 10 requests per day. This resets at midnight UTC. Your remaining requests are shown in the navigation bar after your first query.
| Tier | Requests/day | Cost |
|---|---|---|
| Free | 10 | Free |
| Research (coming soon) | Unlimited | Contact us |
Need higher limits? Email support@peptideagent.xyz.
Sequence Generation
Sequences are generated by GPT-4o with a domain-specific system prompt trained on peptide biology principles. The model outputs single-letter amino acid codes using the standard 20 canonical amino acids:
Generated sequences are typically 6–20 amino acids in length, optimized for the described target based on known structural motifs (amphipathicity, charge distribution, hydrophobicity patterns).
Output Fields
| Field | Description |
|---|---|
sequence | The suggested amino acid sequence in single-letter code |
confidence | AI confidence score (60–95%) based on query clarity and known data |
explanation | Scientific rationale for the sequence design |
mechanism | Predicted biological mechanism tag |
stability | Predicted stability: High / Medium / Low |
halfLife | Estimated half-life in biological media |
toxicity | Predicted toxicity flag (non-toxic / low / moderate) |
molecularWeight | Approximate molecular weight in kDa |
similarity | Top 3 similar known peptides from UniProt |
Similarity Search
After generating a sequence, the platform queries the UniProt REST API to find structurally or functionally similar known peptides. This provides biological context and validation anchors for the generated candidate.
If UniProt returns no close matches, a curated fallback set of well-known reference peptides is used based on the predicted mechanism.
Supported Conditions
The system accepts any plain-English description of a biological target or condition. Pre-built examples include:
You can also enter custom targets like receptor names, pathogen names, or specific biological pathways.
Mechanism Tags
| Tag | Description |
|---|---|
Antimicrobial | Disrupts bacterial or fungal cell membranes |
Anti-inflammatory | Modulates inflammatory cytokine pathways |
Hormonal | Mimics or modulates hormonal signaling (e.g. GLP-1, insulin) |
Neuropeptide | Acts on neural signaling pathways |
Antifungal | Targets fungal cell wall or membrane |
Antiviral | Inhibits viral entry or replication |
Cell-signaling | Modulates intracellular signaling cascades |
Wound-healing | Promotes tissue repair and regeneration |
Disclaimer
For questions, contact support@peptideagent.xyz.