Guy Durant

DPhil Student at University of Oxford. Interested in biology, chemistry, computer science and politics!

[Oxford Aussie Rules @OUARFC]

💙OXFORD Men win their match!

Oxford Men 12.11.83
Cambridge 7.14.56

[Guy Durant @GuyDurant2]

If you're at #AMLD and curious about how ML is transforming drug discovery, swing by!

[Guy Durant @GuyDurant2]

🚀 Excited to speak at AMLD on Data Drives Discovery: The Future of Machine Learning for Small-Molecule Drug Discovery.

📅 Date: 13th February
⏰ Time: 2:10 PM
📍 At SwissTech Convention Center (hosted by Cradle & Adaptyv)

@cradlebio @adaptyvbio @appliedmldays

[Guy Durant @GuyDurant2]

Excited to share our perspective article in Nature Computational Science!

We explore how ML can transform small-molecule drug discovery, but holistically addressing data quality, diversity, and validation is key to breakthroughs.

#AI #DrugDiscovery #ML
https://t.co/q5urgcjHEd

[Nature Computational Science @NatComputSci]

📢Out now! Charlotte M. Deane and colleagues discuss the need for a renewed focus on data and validation for advancing the potential of machine learning in small-molecule drug discovery. @OPIGlets https://t.co/qSS9J7EnUy

➡️https://t.co/lwxHhbo2h9

[Eve Richardson @EveRichardson20]

Read our (@bpeters) article in Patterns explaining why contrary to recent popular belief, the ROC-AUC does not inflate performance estimates in imbalanced data 👇

[Australian Embassy, Croatia @AusAmbCroatia]

It felt like a Saturday afternoon in suburban Australia when the Oxford Uni AFL club took on local AFL teams in Zagreb. Well done to all involved. Football was the winner 🏉🇦🇺🇭🇷🥇@OUARFC @AFLEurope

[Oxford Protein Informatics Group (OPIG) @OPIGlets]

(1/2) The webserver and GitHub for AntiFold, our antibody inverse folding model developed by @magnushoie and @AlissaHummer, have now been released!

Webserver: https://t.co/TIqyGrboVE

GitHub: https://t.co/P0f5nMyPfe

[Keith Hornberger @KRHornberger]

This @cenmag article discussing the SaaS-ification of AI services in drug discovery is interesting, but imo as many of these pieces do, fails to ask the difficult questions. 1/
https://t.co/ig7u7AHW9E

[Oxford Protein Informatics Group (OPIG) @OPIGlets]

OPIG DPhil @YaeliZiv introduces MolSnapper: by integrating expert knowledge into diffusion models for structure-based drug design, we generate similar ligands to the ground truth, while doubling valid molecule output compared to alternative methods.

https://t.co/jObYzd5QkM

[Matt Raybould @mijr12]

My colleage @AlissaHummer & I are organising a conference on Adaptive Immune Receptors: modelling and immunoinformatics, taking place this April 5th at Oxford University.

Invited Talks - Posters (+selected talks) - Networking

More info and registration: https://t.co/JbttNwVcTJ

[AlexGW @ALGW71]

Really happy to share our new study: Investigating the ability of deep learning-based structure prediction to extrapolate and/or enrich the set of antibody CDR canonical forms.

1/N

[Leo Klarner @leoklarner]

Generative models for scientific data can often benefit from using domain-informed constraints to reduce their problem space.
In our new #NeurIPS2023 paper, we simplify constrained (Riemannian) diffusion models to make them better, faster & easier to use.

https://t.co/reCze4g1Pe

[Oxford Protein Informatics Group (OPIG) @OPIGlets]

In our latest preprint, @OPIGlets DPhil student @GuyDurant2 et al. benchmark machine learning-based scoring functions and release a new toolkit (ToolBoxSF) to robustly interrogate performance & determine the effect of dataset biases on predictions
https://t.co/qIlBbyySFT

[Digital Discovery @digital_rsc]

The Academia Prize at @RSC_BMCS @RSC_CICAG #AIChem23, co-sponsored by @PCCP and @RSC_MolEng, was won by Guy Durant (@guydurant2), for "Robustly interrogating machine learning scoring functions: do they learn any biophysics?" Congratulations Guy!

[Oxford Protein Informatics Group (OPIG) @OPIGlets]

New small molecules preprint: Buttenschoen et al. present PoseBusters, a python package for measuring the physical plausibility of docked poses, and apply it as an additional quality control metric in evaluating a suite of recent docking methods.

https://t.co/1zcPQaARRE

[Leo Klarner @leoklarner]

Can we use scientific prior knowledge to make models more robust in low-data, OOD drug discovery?

In our new #ICML2023 paper, we use prior knlg of drug-like chemical space to regularize the function space of neural networks, improving OOD generalization.

https://t.co/AHZbCkmf1B

[Oxford Protein Informatics Group (OPIG) @OPIGlets]

Our most recent immunoinformatics preprint details improvements to our computational epitope binning software (SPACE2): we use the latest antibody 3D modeling technology & benchmark clustering on more precise datasets. Led by @fspoen in collab w/ @Roche

https://t.co/KBZvqIKEGi

[Oxford Tube @Oxford_Tube]

We’re making changes to our fares from Sunday 25th June. This is due to the rising cost inflation, affecting factors such as fuel and engineering parts. No one likes it when prices go up, so we've done our best to keep the increases as low as possible. >https://t.co/iT2jRwmFAI

[Oxford Tube @Oxford_Tube]

Take the kids on a shrek-tastic day out and head to Shrek’s Adventure via the Oxford Tube. Take the Circle or District line from our Victoria stop for a fun-filled interactive experience with the kids this half term.> https://t.co/Heki0cJmCM