pjballester.bsky.social

Struggling to tell which binding affinity prediction methods actually work and which just make the most noise? Our new Nature Machine Intelligence paper offers 7 essential guidelines to raise the bar: rdcu.be/tCU9tkxxqEJa Less hype. More reliable science. #DrugDiscovery #AI

www.linkedin.com/pulse/what-h...

📢 Postdoc position in AI for Antifungal Drug Discovery available at my group 🤝 In collaboration with other groups at Imperial, GSK & Dundee Drug Discovery Unit 📅 Applications close 28 Oct 2026 🔗 All details can be found at careers.iscb.org/jobs/view/10...

A pleasure to attend a stimulating Research Professors networking event at the Royal Society. Excellent discussions on timely topics, followed by a great drinks reception and formal dinner. Many thanks to Sir David Baulcombe and Sir Paul Nurse for hosting and for the kind invitation.

Check out📖https://doi.org/10.1039/D6DD00399K 🚀 TRPM8 virtual screening study. Inactive-enriched structure-based ML shows top performance on novel chemical space. Without requiring 3D structures of this target, fine-tuning BIND, a structure-free protein language model, strong alternative.

🧬 New paper: AI-driven drug repositioning using complementary #MachineLearning models achieved an 82% prospective hit rate and picomolar potency. A promising approach to fight #AntimicrobialResistance 📄 advanced.onlinelibrary.wiley.com/doi/10.1002/...

Check our new paper link.springer.com/article/10.1... If you think that off-the-shelf end-to-end deep-learning docking or protein–ligand co-folding methods are the best approaches for virtual screening, here you have a reproducible counterexample.

Interested in gaining a PhD at the interface between AI and structure-based drug design? A fully funded PhD studentship is available at my group at Imperial College London www.linkedin.com/pulse/phd-st...

Enjoyed speaking at the SAE Media Group's AI in Drug Discovery conference. A big thank you to them for organising and also for my interview www.smgconferences.com/documentport...

Enjoyed reading "Hit identification in ultra large virtual screening: an integrative review and future challenges" www.sciencedirect.com/science/arti...

Interested in drugging ion channels? Check out our study www.frontiersin.org/journals/dru...

So much done, so much more left to do: "Seven in 10 people now survive their cancer five years or more, up from only half in the mid-70s" pressroom.cancer.org/cancer-stati...

FutureHouse has announced the 2026 AI-for-Science Independent Postdoctoral Fellowship (deadline 13 Feb 2026). Applicants need a co-advisor. I am happy to consider candidates: email p.ballester@imperial.ac.uk with CV + 2-page proposal. All info at www.futurehouse.org/fellowship

Looking forward to speaking at this AI for Drug Discovery conference www.smgconferences.com/pharmaceutic...

Brandolini’s Bullshit Asymmetry Principle: “The amount of energy needed to refute bullshit is an order of magnitude bigger than to produce it.” And this is often what it means to be a referee nowadays...

Happy to share our latest research on predicting temozolomide response in low-grade glioma patients using large-scale machine learning. We explored 12 machine learning classification algorithms across six types of omics data. link.springer.com/article/10.1...

🚀 Just published in Pattern Recognition: We present LDMO-CCP, a conformal prediction model using molecular clustering for better uncertainty quantification across chemical spaces. 🔹 Identifies Proscilladin as broad-spectrum cell-active inhibitor 👉 doi.org/10.1016/j.pa...

Really enjoyed my visit and talk last week at UC Berkeley (bidmap.berkeley.edu/seminars/ped...). Many thanks to BidMap for the warm hospitality!