---name: protein-structure-prediction description: Predicts 3D protein structures from amino acid sequences using ESMFold or AlphaFold3 (mock). license: MIT metadata: author: AI Group version: "1.0.0" compatibility:

• system: Python 3.10+ allowed-tools:
• run_shell_command
• read_file

keywords:

• protein-structure
• automation
• biomedical measurable_outcome: execute task with >95% success rate. ---"

Protein Structure Prediction (ESMFold/AF3)

The Protein Structure Prediction Skill provides an interface to state-of-the-art folding models. It takes an amino acid sequence and returns a PDB file or structure metrics (pLDDT).

When to Use This Skill

• When you have a protein sequence and need its 3D coordinates.
• To check if a designed sequence folds into a stable structure.
• To prepare a receptor for docking simulations.

Core Capabilities

1. Folding: Generates atomic coordinates (PDB format).
2. Confidence Scoring: Returns pLDDT scores per residue.
3. Visualization: (Optional) Generates a static view of the structure.

Workflow

1. Input: Amino acid sequence (FASTA string).
2. Process: Sends sequence to ESMFold API (or local inference).
3. Output: Saves .pdb file and returns confidence metrics.

Example Usage

User: "Fold this sequence: MKTIIALSY..."

Agent Action:

python3 Skills/Drug_Discovery/Protein_Structure/esmfold_client.py \
    --sequence "MKTIIALSYIFCLVFDYDY" \
    --output structure.pdb