The Q4Proteins team focuses on large, mostly weakly correlated systems, which will drive the research on quantum simulation of bimolecular systems into a new regime with potentially reduced circuit depth but larger qubit numbers, significantly extending the traditionally limited focus on small, highly correlated systems. We are developing a universal quantum-driven atomistic simulation approach for biochemical problems of importance to human health.
The Q4Proteins team consists of 4 PIs:
and their co-workers, which comprise so far:
We focus on interactions within and between biomolecules and small molecules. Challenging examples include the individually weak interactions between biomolecules such as those found in biomolecular condensates. These are difficult to approach by traditional simulations since a reliable description of the balance of many weak interactions is hard to establish and because they are highly dynamic.
Since quantum computations will be inherently limited to quantum regions involving only a fraction of
atoms of the total system, we will develop a reliable and robust quantum-in-quantum embedding approach with
multiple subsystems. The quantum subsystem data will be used as input to a machine-learned force field that
will eventually allow us to propagate the system in time, necessary for proper sampling of the configuration
space. Our work builds upon expertise that we gained,
together with others, in the Q4Bio competition of Wellcome Leap. Our new approach to the problem will culminate
into the extended pipeline FreeQuantumX. The generality of our approach will enable us to couple our methodology
to existing rare-event sampling methods.
The quantum computation of electronic energies for the inner quantum regions in the embedding scheme will be
carried out by quantum algorithms with rigorous error control to obtain subsystem energies (and possibly forces)
with controllable accuracy. Machine learning will tie different data together. These developments will enable us
to deliver a simulation pipeline that can be applied to large biochemical problems out of the box. Due to its
first-principles character, it will not be restricted to specific molecule classes and it can be applied to
different tasks ranging from drug discovery and design to elucidating complex biomolecular machinery.