QFoldAgent: An Autonomous Quantum Optimization Multi-Agent System for Protein Structure Prediction

Short summary

QFoldAgent is a closed-loop multi-agent framework for quantum-classical protein structure prediction on a 5-residue tetrahedral lattice. A design agent proposes sequence-conditioned penalties, a VQE pipeline optimizes the Hamiltonian under Qiskit Aer noise, and a feedback agent refines penalties using energy-landscape diagnostics. On QDockBank benchmarks it reduces median RMSD from 3.64 to 3.20 Å and raises structural validity on unseen sequences from 87.5% to 98.7%.

• •QFoldAgent uses a multi-agent loop for quantum-classical protein folding on a tetrahedral lattice
• •Reduces median RMSD from 3.64 to 3.20 Å on QDockBank benchmark
• •Raises structural validity from 87.5% to 98.7% on unseen sequences

Generated with AI, which can make mistakes.