Macromolecular modelling algorithms harnessing quantum computation

Proteins, ribonucleic acids, peptides, and synthetic heteropolymers can fold into intricate three-dimensional structures determined by the primary sequence of building-blocks, and can self-assemble into larger multimolecular complexes. The folds and assemblies of macromolecules in turn determine their functions. Although major advances have been made in modelling the sequence-fold-function relationship on classical computers, these are plagued by the vastness of both conformational and sequence spaces accessible to macromolecules. Where classical computers must consider conformations or sequences by iteration over many possibilities, quantum computers have the potential to consider vast numbers of conformations or sequences simultaneously, by implicitly representing these as a superposition of quantum states, and to allow efficient sampling from the low-energy conformations or sequences. This presentation reviews our past work showing that heteropolymer sequences can be robustly designed using adiabatic quantum annealers, and that the design problem can also be mapped to gate-based quantum computers. Here, we also introduce a method for tackling the multimolecular docking problem on adiabatic quantum annealers, without coarsening the molecular representation or simplifying the problem. This problem, which is central both to drug design and validation pipelines and to peptide and protein structure prediction (particularly in the context of solvent molecules), has a solution space that scales exponentially with the number of bodies to be docked, which means that it grows rapidly intractable by classical algorithms. The mapping to quantum computers provides a path to tackling currently intractable multibody docking problems as quantum hardware matures, and could one day greatly accelerate peptide and protein drug development.