QFold: quantum walks and deep learning to solve protein folding

Abstract We develop quantum computational tools to predict the 3D structure of proteins, one most important problems in current biochemical research. explain how combine recent deep learning advances with well known technique walks applied a Metropolis algorithm. The result, QFold, is fully scalable hybrid algorithm that, contrast previous approaches, does not require lattice model simplification and instead relies on much more realistic assumption parameterization terms torsion angles amino acids. compare it its classical analog for different annealing schedules find polynomial advantage, implement minimal realization IBMQ Casablanca system.