【Folding@Home】なぜコロナ研究にスパコンレベルの処理能力が必要なのか？

The Voelz and Chodera labs have been working with an international team of researchers to computationally screen potential inhibitors of the SARS-CoV-2 coronavirus main protease (also known as SARS-CoV-2 MPro or 3CL). After the virus gets into a cell, it co-opts the cell’s translational machinery and makes a polyprotein (one long chain of amino acids) that needs to be cleaved into smaller pieces that become the functional parts of the virus. The main protease is the viral protein that does the cleavage.
Timely high-throughput crystallization work by the Diamond Light Source XChem project has identified drug fragments that bind the protease, and now the race is on to use these initial hits to drive computational prioritization of compounds to synthesize. We are using the power of Folding@home — now the largest supercomputer in the world — to screen tens of thousands of promising molecules that have been computationally docked to the protease. These include over 6000 compounds designed and submitted as part of the COVID Moonshot challenge.

About the calculations

We are using an alchemical free energy calculation method called free energy perturbation, or FEP, to estimate the affinity of a potential drug molecule to its receptor. FEP methods are very expensive compared to other methods like docking, but with the advantages of high accuracy and physical insight. We are using protocols very similar to those used in the most recent SAMPL blind challenge.
The calculation involves decoupling the drug molecule (by turning off its electrostatic and dispersion interactions) from its receptor, and again for a molecule in water. For the latter type of simulations, unfortunately there’s not much to look at the Folding@home viewer, because there’s no protein being simulated. However boring these projects appear, please keep in mind that they are absolutely essential to our success.