Dr Liam Brierley

A collage promoting Bright Club Glasgow. The top left panel features a cartoon blue rubber duck wearing a red-and-white traffic cone as a party hat, with the text "Bright Club Glasgow" on its side. Surrounding this image are five photos of individual stand-up performers on stage with microphones, each speaking animatedly or smiling

Bright Club is back! 

Join us at The Stand Glasgow for a night of laughs as our researchers and academics turn into stand-up comedians.

📅 Mon 10 November 
📍 The Stand, Glasgow 

Get your tickets for £4 using discount code "BRIGHTCLUB": gla.ac/4hivOMZ

You can when smartass titles become a tickbox in the CRediT author taxonomy 😁

This is the 2nd preprint to come from collaboration between @thepandemicinst.bsky.social @livuniresearch.bsky.social & CSL Seqirus, supported by @baylism.bsky.social & Joaquin Mould.

We hope this work can help risk assess new bird flu strains and flag key mutations in the wild!

#preprint #avianflu

A dotplot showing predicted zoonotic probability for all test sequences, which represent 13 subtypes in total. There are blue bird-only sequences toward the left, which generally cluster below the threshold for zoonotic prediction, and red confirmed zoonotic sequences on the right which show subtype specific patterns. Human zoonotic H5N1 sequences are predicted with consistent confidence, all just over the threshold to be considered zoonotic, while H7N9 shows much more variability with a fraction of sequences being misclassified as not zoonotic, and a fraction being classified correctly with extremely strong confidence. Of note is that most sequences of rare spillover events (<10 sequences available, like H10N8 and H3N8) were correctly classified with middling confidence. Highlighted and annotated are also three bird origin sequences that are distinctly higher in predicted zoonotic risk than all other sequences of their same subtypes. They are: H4N6 A/American black duck/New Brunswick/00499/2010, H4N6 A/yellow-billed teal/Argentina/CIP051-91/2011, and H4N8 A/American black duck/New Brunswick/02375/2007.

This stack is able to correctly predict zoonotic potential of sequences in entirely unseen subtypes with AUC=0.95 and F1=0.90, a level of generalisability that is not often seen for machine learning host predictors.

Interestingly, it flags some duck H4 viruses from Americas as having distinct risk.

A heatmap showing how well each modelled combination of feature set and gene/protein predicted zoonotic potential. Protein features showed slightly higher AUC across the board, while the strongest gene/proteins for zoonotic prediction were PB2, PB1, HA, NP, and NS1. We tried five different algorithms, and each was kept at least once, but the best performing algorithms were usually RF (random forests) or XGB (XGBoost).

Training on 12 feature sets over each of 8 segments, we find protein properties are usually best at estimating zoonotic potential from a single segment.

But what about whole genomes? We can combine the best models in a single trained meta-learner (or "stack"), that draws on info from all of them!

Histogram over time of the most common avian flu subtypes in our dataset by sequence frequency. Most zoonotic sequences are from H5N1 and H7N9, and peak around 2005-2015. Avian sequences are more mixed and feature many more subtypes including H9N2, H3N8, H10N7, and H5N1, all at highly variable frequencies over time.

We extracted ~19000 influenza sequences from birds and ~600 zoonotic sequences from humans (only non-seasonal subtypes).

Before training, we remove redundancy by grouping similar sequences into clusters. This is important to reduce bias, as most come from just a few subtypes like H7N9 and H5N1.

Model training diagram emphasising we hold out an entire subtype of avian influenza during training in order to make unbiased predictions on it, and that we use multiple algorithms, multiple feature sets, and all eight segments to train models

Lots of ML models can predict human spillover. However for influenza this task is harder because of a) genome segmentation, and b) strong signal within subtype or lineage.

We planned a model training architecture to handle this, ensuring predictions are rooted in virus biology, not shared ancestry.

An AI for an AI: identifying zoonotic potential of avian influenza viruses via genomic machine learning Avian influenza remains a serious risk to human health via zoonotic transmission, as well as a feasible pandemic threat. Although limited zoonotic cases have resulted from the current epizootic outbre...

🚨📝 New work via @thepandemicinst.bsky.social now preprinted!

We took ~19000 sequences of avian influenza virus and trained machine learning models on >40000 genomic/proteomic features - the resulting ensemble can predict zoonotic potential with AUC=0.95. More below!

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

Experimental infections reveal unexceptional viral tolerance in bats https://www.biorxiv.org/content/10.1101/2025.08.28.672855v1

That'll be St Elmo - same chap who gives his name to the nautical fire!

Virus researchers! Please consider participating in this project - it would be a huge help to our lab, and we think it'll lead to some really exciting synthesis. Plus, you'll get an invitation to participate in a workshop later in the project! 🦠😷

We have a job going for a bioinformatician / computational biologist at the CVR in Glasgow. www.jobs.gla.ac.uk/job/computat...

What is this nightmare of flesh

If I had a nickel for every consecutive year we've had janky 3D-rendered big cats, I'd only have two nickels but it's weird it's happened twice.

A protein language model for exploring viral fitness landscapes - Nature Communications Ito et al. present CoVFit, an AI model that predicts variant fitness (transmissibility) from spike protein sequences alone. They further demonstrate its utility in forecasting viral evolution via sing...

I look forward to reading www.nature.com/articles/s41...!

I'm only able to join ViBioM virtually (thank you @evbc.bsky.social for making this available!!) but some fantastic talks from the @systemsvirology.bsky.social lab who are innovately retraining protein language models to infer virus evolution, antigenicity, and epi! おめでとうございます

@uofgpe.bsky.social
@uofglasgow.bsky.social
@unistrathclyde.bsky.social

So much #scicomm in Glasgow in the next month!

Talks in a pub? ✔️ - @pintofscience.uk: pintofscience.co.uk/events/glasgow

Family activities? - ✔️ Glasgow Science Festival: www.gla.ac.uk/events/scien...

and my (biased) fave - Science comedy??? - ✔️ Bright Club: www.thestand.co.uk/performance/...

(I know it's the same guy cause the photocopy always caught his shirt cuff in the scan)

I was tracing the first human reports of every RNA virus in my PhD and I quickly learned about interlibrary loans!

Any uni library can make a request for a paper for you, usually I'd get pdfs, sometimes hardcopies, sometimes even photocopies from a guy at the British library who wore funky shirts.

Last day for abstract submission to present your work at the 𝗚𝗹𝗮𝘀𝗴𝗼𝘄 𝗩𝗶𝗿𝗼𝗹𝗼𝗴𝘆 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽 and connect with researchers across all directions of virology! We would love to have more representation and contributions from ECRs and students!

Abstracts info in link below!

🎓Big Congratulations to Dr Hollie French for passing her Viva yesterday! 🎉

📄Thesis: ‘Genomic Surveillance and Biogeography of Vampire Bat Rabies in Central America and Mexico.’ 🦇🦠and thank you to her supervisor:
@danielstreicker.bsky.social

Yes please! Great to see!

Influenza of avian origin confirmed in a sheep in Yorkshire Influenza of avian origin (H5N1) has been confirmed in a single sheep in Yorkshire.

The UK’s Chief Veterinary Officer has confirmed a case of highly pathogenic avian influenza (H5N1) in a single sheep in Yorkshire following repeat positive milk testing.

To my knowledge, this is the first time the virus has been detected in sheep.

www.gov.uk/government/n...

Comparative study of tropism and emergence of virulent companion animal viruses using AI-powered text mining at University of Glasgow on FindAPhD.com PhD Project - Comparative study of tropism and emergence of virulent companion animal viruses using AI-powered text mining at University of Glasgow, listed on FindAPhD.com

Come do a PhD with me using LLMs and comparative methods to understand the emergence of companion animal viruses! 🧠☠️🕸️🐕🐈🖥️

Fully funded 3.5 year PhD based at @sbohvm.gla.ac.uk and @cvrinfo.bsky.social supervised by myself, Margaret Hosie, and Willie Weir

www.findaphd.com/phds/project...

We love to hear about about the amazing work of our William Guy Lecturers in bringing stats to young people!

Want to inspire the next generation with a talk about stats and AI?

There's still time to apply for our 2025/26 lectureship 👉 rss.org.uk/news-publica...

Don't forget this is tonight! Last tickets available below!

yes yes yes yes yes yes!!

Get them tickets if you want some dazzling academics tickling your funny bones with tales of asexuality, booze, saints, sensory stimulation, and more!

www.thestand.co.uk/performance/...

🎺 PhD studentship available at @lshtm.bsky.social and RVC! @influenzal.bsky.social and I are advertising for a project to investigate the unexpected phenomenon that mutagenic antiviral drugs can give rise to viruses with hundreds of mutations which are still viable. 🧵
www.lshtm.ac.uk/study/fees-a...

For those aiming to go that route, what's the one thing you would change about how postdoc interviews work?

asking b/c we're hiring and I am all about making that interview process equitable and accessible if I can!