[CINECA - PoliMi] Molecular unfolding with Quantum Annealing
Hello everybody. I am Daniele Ottaviani, I write on behalf of the joint team "CINECA - Politecnico di Milano".
For a long time, CINECA and PoliMi have been dealing with a molecular docking application, being able to exploit the powerful computational resources available to us (for those who don't know, CINECA is the largest Italian supercomputing center and owner of "Marconi", a supercomputer in nineteenth place in the top500 list).
CINECA is currently involved in the front line fighting against COVID-19 through the Exscalate4CoV project. In this first phase, we are simulating the behavior of proteins that allow the virus to replicate so that we can virtually test the most effective pharmaceutical molecules to inhibit the virus, and then we can go to the validation phase in the laboratory by accelerating the production of drugs effective to reduce its replicability. Each protein requires at least one week of continuous simulation on 16 nodes of the Cineca supercomputer. With a normal computer, it would take at least 4 months for each protein. The results of the simulations will then be analyzed via the Exscalate platform.
Thanks to our previous knowledge, we are therefore able to access a vast database that can be used for this project. Among the various problems that can be solved with a Quantum Annealer, we have chosen one in particular: the molecular unfolding problem.
By molecular unfolding we mean the preliminary process to the docking phase, where the molecules considered "suitable" for experimentation are manipulated so that the distance between its atoms is as large as possible. Currently, this operation is carried out through the use of empirical techniques that do not always allow us to find the best molecular conformation, compromising the quality of the docking.
The reduced size of the molecules to be tested and the objective that requires a great quality of the solution make the problem potentially feasible to solve through the use of the Quantum Annealer D-WAVE.
At the moment we are still in the writing phase of the QUBO problem. Any suggestions or requests for clarifications are welcome!
Regards,
Daniele
Comments
Hi Daniele,
Is unfolding in this context like running MD at a higher temperature? Are the empirical techniques like this? Just trying to understand.. If the number of molecules is an issue, you can always substitute our HSS service for the QPU.
Thanks,
Marcus
Hi Marcus,
Our problem is the following: given a molecule with rotatable bonds, we want to find the ”unfolded” torsion configuration that maximizes the distances between atoms. We think that this paper could be used as a basis to solve our problem.
Depending on the number on the number of atoms (or rigid bodies) and rotatable bonds considered we will try both QPU and HSS.
Thanks,
Riccardo
Please sign in to leave a comment.