On a separate note: @klyneai.bsky.social develops practical, affordable software for drug discovery! Reach out at [[email protected]]. Our team supports all stages with a pay-as-you-go model, no licensing fees, and costs nearly an order of magnitude lower than competitors—while maintaining quality 😊

Happy to answer any questions about this! 🚀

A pleasure working with so many awesome people @Mohammad Ghazi Vakili, @cgorgulla.bsky.social , @aspuru.bsky.social, Igor Stagljar & @elonverse.bsky.social e.bsky.social & so many more!

3️⃣ What’s next? This is just the beginning. Generative models have immense potential to accelerate drug discovery. Future work will explore larger quantum systems and transformer-based models. Though not yet practical, advancements in quantum technologies will greatly enhance their impact.

2️⃣ Quantum vs. classical approaches: Could classical methods achieve similar results (e.g., Tartarus benchmarking)? Probably—but we observed something fascinating. More qubits in the quantum model improved its ability to generate stable, synthesizable molecules, suggesting better distb. modeling:

1️⃣ Early-stage inhibitors, not yet drugs: The inhibitors we identified are still in the early stages of development—not ready to be called "drugs." However, this work demonstrates the potential of quantum generative models in molecule design and their applicability in real-world scenarios.

4️⃣ Quantum computing + KRAS inhibitors: arxiv.org/abs/2402.08210

Full list of publications: scholar.google.com/citations?hl...

Huge thanks to my advisors, mentors & collaborators for their support! 🚀 #PhDDefense #Stanford #MachineLearning

1️⃣ Designing transcriptional repressors with deep learning: biorxiv.org/content/10.1...
2️⃣ VirtualFlow 2.0: Screening 69B molecules: biorxiv.org/content/10.1...
3️⃣ Tartarus: Benchmarking inverse molecular design: proceedings.neurips.cc/paper_files/...

Thank you Martin! 🙂

Thanks Kjell! 😄