Toni Mey

Benchmarking Active Learning Protocols for Ligand-Binding Affinity Prediction Active learning (AL) has become a powerful tool in computational drug discovery, enabling the identification of top binders from vast molecular libraries. To design a robust AL protocol, it is important to understand the influence of AL parameters, as well as the features of the data sets on the outcomes. We use four affinity data sets for different targets (TYK2, USP7, D2R, Mpro) to systematically evaluate the performance of machine learning models [Gaussian process (GP) model and Chemprop model], sample selection protocols, and the batch size based on metrics describing the overall predictive power of the model (R2, Spearman rank, root-mean-square error) as well as the accurate identification of top 2%/5% binders (Recall, F1 score). Both models have a comparable Recall of top binders on large data sets, but the GP model surpasses the Chemprop model when training data are sparse. A larger initial batch size, especially on diverse data sets, increased the Recall of both models as well as overall correlation metrics. However, for subsequent cycles, smaller batch sizes of 20 or 30 compounds proved to be desirable. Furthermore, adding artificial Gaussian noise to the data up to a certain threshold still allowed the model to identify clusters with top-scoring compounds. However, excessive noise (<1σ) did impact the model’s predictive and exploitative capabilities.

Rohan has done fantastic work on machine learning for binding affinity predictions and active learning. Check out his papers here: pubs.acs.org/doi/full/10....

pubs.acs.org/doi/full/10....

and

www.biorxiv.org/content/10.1...

Congratulations to Dr @rohangorantla.bsky.social for passing his PhD viva today marking the first official Mey group Dr. Thank you to Andrea Volkamer and Chris Wood for conducting the viva and all the best on your future scientific adventures Rohan!

163163 Associate professor or Tenure-track assistant professor in computational drug development

Associate professor or Tenure-track assistant professor in computational drug development, U Copenhagen 🇩🇰 employment.ku.dk/faculty/?sho... #compchem

JOSS if the software is open source!

Looks like an RDkit visualisation?

I have definitely suggested to authors of papers I have reviewed to find a chemist to talk to before presenting molecules.

Probably part of some conference proceeding?

Seems the issue is there are two different objectives. 1. DL community wants to show they are better than a not very well thought out benchmark, often not with basic significance testing on bold numbers 2. Users realises the DL benchmark has nothing to do with the use case.

Wow very impressive! Congratulations 🎉. Looking forward to the code ☺️

Check out @jjguven.bsky.social work on free energy calculations on zinc-based metalloenzyme free energy calculations #chemsky

Photograph of Johannes Margraph and Günter Klambauer introducing the ELLIS ML4Molecules Workshop 2024 in Berlin at the Fritz-Haber Institute in Dahlem.

Excited to speak at the ELLIS ML4Molecules Workshop 2024 in Berlin!

moleculediscovery.github.io/workshop2024/

Super excited to preprint our work on developing a Biomolecular Emulator (BioEmu): Scalable emulation of protein equilibrium ensembles with generative deep learning from @msftresearch.bsky.social ch AI for Science.

www.biorxiv.org/content/10.1...

📢 It's our great pleasure to invite you all to our upcoming #Workshop on "Computer Simulation and #Theory of Macromolecules" in #Hünfeld on March 7-8, 2025 in #hybrid format! See you online or on-site! More details are provided here: www.mpinat.mpg.de/workshop/hue... #TheoryHünfeld2025 #biophysics

The 4th EFMC2 tandem talk 5th Dec (tomorrow) at 4pm CEST. Register ➡️ www.efmctandemtalks.org. Taylor Quinn and colleagues from AstraZeneca will present work on "Accelerated Discovery of a Carbamate Scaffold Cbl-b Inhibitor Using Generative Models and Structure-based Drug Design".

I am almost disappointed now

Wait what?

Yep and I keep blocking them and for a brief period tried the ‘unsubscribe’ option, just wondering how I could teach my spam filter that I don’t want these!

Does anyone have any strategies for unsolicited emails from irrelevant publishers that are clogging up an inbox? My research really is not suitable for an astrophysics and astronomy journal even if you send me 4+ emails per day! #chemsky

Off to Berlin for a week for hopefully lots of fun science!

People in a university room with laptops and someone playing a guitar

People in a university room presenting data.

Presenting the concept of the datajam. People are in a room looking at a presentation.

Yesterday we had a data jam at UoE Chemistry to turn data into music. We'll be showcasing the final pieces in January so watch this space in the meantime some fun pictures from the event.

I discovered the 3D printer at the university library. All the supports seemed a lot easier to remove in my head…

First snow this year! Just in time for talking about thermodynamics of phase changes in lectures.

The Small Molecule Digital Chemistry group at Novo Nordisk is looking for a Senior Scientist in the computational chemistry team. Join a young, dynamic team of drug discovery scientists working with an in silico first mindset to advance pharmaceutical programs! *Closing date Nov 30th
#chemjob

Looking fwd to see @altmetric.com linking @bsky.app to bioRxiv @biorxivpreprint.bsky.social and to all major journals that are in here or about to land! 👏🏻🤩💙🦋

If there is still room I would love to be added!

You have joined!

Hi Bluesky! 👋
I'm a PhD student working on ML methods for Drug Discovery at Edinburgh with @ppxasjsm.bsky.social. Looking forward to connecting with fellow ML, CompChem, Bio and Drug Discovery folks!🧬💊💻

Sharing starter packs in 🧵 below that I found helpful in finding people working in the domain.

New preprint! We prospectively evaluate structure-based methods and measure how well they classify clinical kinase mutations as inhibitor–resistant or –sensitizing when compared against our NanoBRET assay. Alchemical methods are well-suited to modeling distal mutants! Link: doi.org/10.1101/2024...

Aren’t many very similar size? Trying to figure out which one is bigger/small would take too much time for my liking! Surely the only way to arrange backs is by colour!

Yesterday’s speaker dinner after a wonderful seminar by Dr Erna Wieduwilt from SDU telling us about LPMOs role in biomass catalysis!