Thread by @nanogenomic

There weren't any SARS-CoV-2 omicron variant spike protein structures online, so I simulated it and docked its binding to ACE2, Vir antibodies, and Regeneron antibodies. Considering publishing but done just for fun at the moment. Please cite if you want to publish with this.

PDB files of omicron spike available here: ligandal.com/uploads/5/7/9/…

Credit to SwissModel and PDB for structures.
Vir antibodies: rcsb.org/structure/6WPS
Regeneron antibodies: rcsb.org/structure/6XDG
SARS-CoV-2 w/ ACE2: rcsb.org/structure/7KJ2

Red amino acids are mutations or deletions (-) in the spike protein colored in red compared to the wildtype, original spike protein of SARS-CoV-2.

Day 2: ran thermodynamic modeling of Vir (chains C and D) Fab antibody docked to the spike protein receptor binding domains of omicron (chain A) vs. wildtype (chain B). The Vir Fab Gibbs free energy (∆G ) decreasing suggesting Vir binds even more strongly to omicron (PDBePISA).

For fun, here's a script that colorizes and extends the atoms at the binding interfaces of the Vir antibodies with the wildtype or omicron variants. You can really zoom in and see the predicted interactions.

The Regeneron antibodies are a pair, so there are more interfaces. A represents the omicron RBD, B represents the wildtype RBD, and chains C, D, E and F are the Fab antibody fragments from the REGN antibody cocktail. The largest surface area interactions seem to weaken.

This suggests that Vir should gain or maintain efficacy, while Regeneron should lose some efficacy. These are crude, computationally minimalistic thermodynamic simulations based on rudimentary dockings. It is not a precise predictive tool at predicting how folding may change.

∆G is a nifty concept. Oil and water will have +∆G, while compounds in a similar state where energy would have to be exerted to separate them will have a -∆G. The more negative the ∆G, the stronger the biomolecular interactions between two binders (or, a ligand and receptor).

A lot of groups focus on computationally intensive simulation techniques. At Ligandal, our thesis has been to keep things simple and minimize computational load in order to get to actual biophysical or wet lab experiments as quickly as possible.

To learn more about how we apply these kinds of approaches to designing synthetic peptide binders against specific markers, see our Peptide Antidotes to SARS-CoV-2 publication from last year where we designed a synthetic peptide as a COVID therapeutic.

biorxiv.org/content/10.110…

Much of our past work has—and future work will—apply to designing ligands to act as predictive targeting molecules for nanoscale delivery systems, expanding the scope and applicability of technologies such as CRISPR, RNA, DNA, and siRNA.

Want to work with us, partner or support our work on our mission of delivering genetic medicine? DMs are always open.