We have published the prelim. speaker lineup and program over at the event page, we're incredibly excited for this lineup (including a keynote from Max Welling @wellingmax)!

🥳🎉 Registration for the Hünfeld #Workshop "Computer Simulation and #Theory of #Macromolecules" is now open! Don't miss your chance to join us! For more details see 🔗 https://t.co/blTupgmCot

Reminder: Our nonprofit, the Open Molecular Software Foundation, is hiring a research software engineer and a technical lead. Please encourage talented science software folks to apply!

Registration to the 2024 Workshop on Free Energy Methods in Drug Design is now open: https://t.co/r050FpycPb

Introducing the organizing committee for #alchemy2024: Barbara Zarzycka, Bert de Groot, David Hahn, Dima Lupyan, @HighSpeedMode, @hugogdtc, @jenkescheen, @VytasGapsys, @Willemjespers

The 2024 edition of the Free Energy Workshop will take place in Leiden, Netherlands, 13-15th May, 2024! We have not yet opened registration, but you can already book the dates in your calendars. Keep an eye on the wiki for further information: https://t.co/OGRFzVzLH7 #alchemy2024

Thanks everyone involved! It was great to work together on this. What we built here is an automated workflow for free energy calculations: put ligand SMILES in, get dG back. With this in hand we probed how various docking protocols would affect dG accuracy

Finally, we tested the approach prospectively and in this computational exercise it looked promising (~3 kcal/mol improvement on the previous best binder). Next is to try it in a real drug discovery project.

We were also able to control the way in which to explore the chemical libraries. By changing ligand selection strategy we could a) quickly find a deep local minimum of strong binders or b) obtain a good overall description of the chemical space

First, we needed to calibrate the active learning protocol: in our hands, predictions based only on simple ligand descriptors outperformed any other fancier type of representation.

Our latest work on alchemical free energies. This time we used them as an oracle in an active learning loop to navigate chemical space in search for high affinity binders. Thanks to Yuriy, @gtresadern, David and Bert! https://t.co/J3qnReB28P Several main findings below

<- Computational Chemist with background in Quantum Chemistry, protein Molecular Dynamics and industrial experience in Material Science searching for a new position. Ideally in the Karlsruhe/Heidelberg area but not limited to it. Find my LinkedIn profile:

Pre-Exascale Computing of Protein–Ligand Binding #FreeEnergies with Open Source Software for #DrugDesign #cheminformatics https://t.co/YZqgwqoyyq @davidlmobley @gtresadern #current_issue #JCIM #Letters

@JCIM_JCTC Thanks for playing. The answer is: 1. The use of maps is not allowed. 2. Clipart is not allowed, unless it is from the Microsoft Suite. I understand that this is probably licensing policy, but feels weird that showing contours of continents is forbidden

Here is something unexpected: just discovered several hidden rules for the @JCIM_JCTC table of contents figure. Can you guess what two rules are violated by this TOC image?

As a result, we also made free energy predictions for the systems in @Ch_Schindler's dataset with the open source force fields, including @openforcefield

It was great to see that the alchemical protocols parallelize really efficiently: we could make use of ~500 nodes of the Max Planck new supercomputer 'Raven' for 3 days (more than 3 million core hours).

Our latest study on ligand alchemy now in @JCIM_ACS with David @gtresadern @davidlmobley Markus and Bert: https://t.co/DEWZZRJLaF The question we asked here: how far can we push the limits of scaling such calculations relying on the open source software and force fields?