Not Banksy, it's based on Loretto but it's been edited.

Cool stuff. Is GOAT reactive? The epoxide is supposed to react with the thiol, right?

Looks cool.

Very informative. Thanks.

Protein–Ligand Interaction Energies from Quantum-Chemical Fragmentation Methods: Upgrading the MFCC-Scheme with Many-Body Contributions Quantum-chemical fragmentation methods offer an attractive approach for the accurate calculation of protein–ligand interaction energies. While the molecular fractionation with conjugate caps (MFCC) sc...

Another recent paper about MBE fragmentation on protein-ligand systems:
pubs.acs.org/doi/10.1021/...

OG

Maybe set enforcePeriodicBox=True when appending the reporter?

Do you think most humans would be able to fix your latex problem?

Good idea, I will check it out. Thanks.

GitHub - choutkaj/GFA-ligand-set: Set of galectin ligands with known affinities. Set of galectin ligands with known affinities. Contribute to choutkaj/GFA-ligand-set development by creating an account on GitHub.

I've recently collated affinities for over 1000 inhibitors of Galectin-1 and -3, all measured by a single assay (fluorescence anisotropy). Ligands are provided as SMILES. Paper is pending.

Anyone is welcome to test their method on this set.

github.com/choutkaj/GFA...

Thanks for this.

Ok, now I get it. Thanks!

I don't see how vacuum would be relevant. The water is not choosing between the binding site and vacuum. It's choosing between the binding site and bulk.

If it's unfavorable (positive deltaG compared to bulk) for the water to be in that position, why it stays there?

Just tried it out. This is next level stuff.

Molecular nodes are great. Might I ask if you plan to add support for double/triple bonds, and delocalized bonds in aromatic rings? Thx

I would say there is a breaking point if the method is so costly that you can only access timescales that do not give you any meaningfull information about the system.

E.g. with QM dynamics, you can maybe simulate a protein for a few ps. That is pretty useless.

For simulating lets say a protein of 2000-5000 atoms in explicit water, more than 100x slowdown will hurt a lot.

Having said that, if the increased cost is paid off by increased accuracy, then there is always a merit.

Would be a funny plottwist if bluesky changed its name to twitter.

Throw in HolyC to spice it up.

This is really great paper. I have been going back to it multiple times. Thanks for all your work.