Robertson Lab
@davidlrobertson.bsky.social
Viruses, evolution, computational biology & other stuff, David L Robertson @robertson_lab (in the other place) based at @cvrinfo.bsky.social, the University of Glasgow.
Pinned
Our PLM-interact is out in Nature Communications! We show that jointly encoding protein pairs using protein language models improves protein–protein interaction prediction performance and enables fine-tuning to predict mutation effects in human PPIs. www.nature.com/articles/s41...
PLM-interact: extending protein language models to predict protein-protein interactions - Nature Communications
Protein structure can be predicted from amino acid sequences with unprecedented accuracy, yet the prediction of protein–protein interactions remains a challenge. Here, authors present a sequence-based...
www.nature.com
New paper👇
Reposted by Robertson Lab
A nice article, highlighting the important work our collaborators in South Africa are doing using wastewater to surveil a bunch of different vaccine preventable diseases (and finding a lot more of them than other reports would make you believe…).
💪🇿🇦
www.nature.com/articles/d44...
💪🇿🇦
www.nature.com/articles/d44...
Wastewater surveillance reveals disease trends in South Africa
From COVID-19 to measles, scientists are showing how wastewater surveillance can expose underreported infections and strengthen national health monitoring.
www.nature.com
November 24, 2025 at 1:28 AM
A nice article, highlighting the important work our collaborators in South Africa are doing using wastewater to surveil a bunch of different vaccine preventable diseases (and finding a lot more of them than other reports would make you believe…).
💪🇿🇦
www.nature.com/articles/d44...
💪🇿🇦
www.nature.com/articles/d44...
Reposted by Robertson Lab
NEW: We traced out a devastating outbreak of bird flu to its source: an egg farm in Ohio. Where the wind blew, the virus followed.
Within weeks, the farms downwind were about 20 times as likely to see outbreaks as those that weren’t.
Within weeks, the farms downwind were about 20 times as likely to see outbreaks as those that weren’t.
What the U.S. Government Is Dismissing That Could Seed a Bird Flu Pandemic
Egg producers suspect bird flu is traveling through the air. After a disastrous Midwestern outbreak early this year, we tested that theory and found that where the wind blew, the virus followed. Vaccines could help, but the USDA hasn’t approved them.
www.propublica.org
November 18, 2025 at 1:00 PM
NEW: We traced out a devastating outbreak of bird flu to its source: an egg farm in Ohio. Where the wind blew, the virus followed.
Within weeks, the farms downwind were about 20 times as likely to see outbreaks as those that weren’t.
Within weeks, the farms downwind were about 20 times as likely to see outbreaks as those that weren’t.
Reposted by Robertson Lab
Human #BirdFlu case in WA. #H5N5 not #H5N1. Patient is severely ill. Update coming.
www.latimes.com/science/stor...
www.latimes.com/science/stor...
Another person in U.S. is hospitalized with bird flu. Officials don't know how they got it
A person in Washington is presumed infected with bird flu virus. Health officials in the state do not know the source.
www.latimes.com
November 14, 2025 at 10:06 PM
Human #BirdFlu case in WA. #H5N5 not #H5N1. Patient is severely ill. Update coming.
www.latimes.com/science/stor...
www.latimes.com/science/stor...
Reposted by Robertson Lab
New paper👇
Our PLM-interact is out in Nature Communications! We show that jointly encoding protein pairs using protein language models improves protein–protein interaction prediction performance and enables fine-tuning to predict mutation effects in human PPIs. www.nature.com/articles/s41...
PLM-interact: extending protein language models to predict protein-protein interactions - Nature Communications
Protein structure can be predicted from amino acid sequences with unprecedented accuracy, yet the prediction of protein–protein interactions remains a challenge. Here, authors present a sequence-based...
www.nature.com
October 28, 2025 at 3:36 PM
New paper👇
Reposted by Robertson Lab
To summarise, PLM-interact extends single-protein PLMs to jointly encode interacting partners. It achieves state-of-the-art performance in cross-species and virus–host PPI prediction tasks and can be fine-tuned to predict mutation effects in human PPIs.
October 28, 2025 at 3:27 PM
To summarise, PLM-interact extends single-protein PLMs to jointly encode interacting partners. It achieves state-of-the-art performance in cross-species and virus–host PPI prediction tasks and can be fine-tuned to predict mutation effects in human PPIs.
Reposted by Robertson Lab
PLM-interact was applied to virus–human PPI prediction. The model outperforms existing approaches, achieving 5.7%, 10.9%, and 11.9% gains in AUPR, F1, and MCC, respectively — effectively capturing virus–host interactions at the protein level.
October 28, 2025 at 3:27 PM
PLM-interact was applied to virus–human PPI prediction. The model outperforms existing approaches, achieving 5.7%, 10.9%, and 11.9% gains in AUPR, F1, and MCC, respectively — effectively capturing virus–host interactions at the protein level.
Reposted by Robertson Lab
We further demonstrate examples where PLM-interact correctly predicts the effects of mutations on PPIs associated with human diseases. These results highlight its potential to identify whether disease-associated mutations weaken or strengthen protein interactions.
October 28, 2025 at 3:27 PM
We further demonstrate examples where PLM-interact correctly predicts the effects of mutations on PPIs associated with human diseases. These results highlight its potential to identify whether disease-associated mutations weaken or strengthen protein interactions.
Reposted by Robertson Lab
We fine-tuned PLM-interact to predict the effects of mutations on protein interactions — identifying whether mutations increase or decrease interaction strength. The fine-tuned model significantly outperforms zero-shot PPI models in the mutation-effect prediction task.
October 28, 2025 at 3:27 PM
We fine-tuned PLM-interact to predict the effects of mutations on protein interactions — identifying whether mutations increase or decrease interaction strength. The fine-tuned model significantly outperforms zero-shot PPI models in the mutation-effect prediction task.
Reposted by Robertson Lab
PLM-interact achieves state-of-the-art performance on a widely adopted cross-species PPI prediction benchmark — trained on human data and tested on mouse, fly, worm, yeast, and E. coli.
October 28, 2025 at 3:27 PM
PLM-interact achieves state-of-the-art performance on a widely adopted cross-species PPI prediction benchmark — trained on human data and tested on mouse, fly, worm, yeast, and E. coli.
Reposted by Robertson Lab
Existing PPI models use pre-trained PLMs to embed each protein separately, ignoring amino acid interactions between proteins. PLM-interact goes beyond single-protein encoding by jointly representing protein pairs to learn their relationships.
October 28, 2025 at 3:27 PM
Existing PPI models use pre-trained PLMs to embed each protein separately, ignoring amino acid interactions between proteins. PLM-interact goes beyond single-protein encoding by jointly representing protein pairs to learn their relationships.
Reposted by Robertson Lab
Protein language models trained on massive protein sequence datasets capture evolutionary, sequence and structural features — becoming the method of choice for representing proteins in state-of-the-art PPI predictors.
October 28, 2025 at 3:27 PM
Protein language models trained on massive protein sequence datasets capture evolutionary, sequence and structural features — becoming the method of choice for representing proteins in state-of-the-art PPI predictors.
Reposted by Robertson Lab
This work is a part of my viroinf PhD research, carried out under the supervision of Craig Macdonald, @davidlrobertson.bsky.social, and Ke Yuan, with HPC support from DiRAC (www.dirac.ac.uk).
October 28, 2025 at 3:27 PM
This work is a part of my viroinf PhD research, carried out under the supervision of Craig Macdonald, @davidlrobertson.bsky.social, and Ke Yuan, with HPC support from DiRAC (www.dirac.ac.uk).
New paper👇
Our PLM-interact is out in Nature Communications! We show that jointly encoding protein pairs using protein language models improves protein–protein interaction prediction performance and enables fine-tuning to predict mutation effects in human PPIs. www.nature.com/articles/s41...
PLM-interact: extending protein language models to predict protein-protein interactions - Nature Communications
Protein structure can be predicted from amino acid sequences with unprecedented accuracy, yet the prediction of protein–protein interactions remains a challenge. Here, authors present a sequence-based...
www.nature.com
October 28, 2025 at 3:36 PM
New paper👇
Reposted by Robertson Lab
A divergent betacoronavirus with a functional furin cleavage site in South American bats https://www.biorxiv.org/content/10.1101/2025.10.24.684489v1
October 27, 2025 at 6:17 PM
A divergent betacoronavirus with a functional furin cleavage site in South American bats https://www.biorxiv.org/content/10.1101/2025.10.24.684489v1
Reposted by Robertson Lab
1. We ( @jbakcoleman.bsky.social, @cailinmeister.bsky.social, @jevinwest.bsky.social, and I) have a new preprint up on the arXiv.
There we explore how social media companies and other online information technology firms are able to manipulate scientific research about the effects of their products.
There we explore how social media companies and other online information technology firms are able to manipulate scientific research about the effects of their products.
October 24, 2025 at 12:47 AM
1. We ( @jbakcoleman.bsky.social, @cailinmeister.bsky.social, @jevinwest.bsky.social, and I) have a new preprint up on the arXiv.
There we explore how social media companies and other online information technology firms are able to manipulate scientific research about the effects of their products.
There we explore how social media companies and other online information technology firms are able to manipulate scientific research about the effects of their products.
Reposted by Robertson Lab
There is one point of agreement among both enthusiasts and sceptics: the future has already arrived
go.nature.com/3WL73iS
go.nature.com/3WL73iS
Universities are embracing AI: will students get smarter or stop thinking?
Millions of students arriving at campuses are now using artificial intelligence. Worries abound.
go.nature.com
October 21, 2025 at 10:34 AM
There is one point of agreement among both enthusiasts and sceptics: the future has already arrived
go.nature.com/3WL73iS
go.nature.com/3WL73iS
Reposted by Robertson Lab
And now NextStrain as well.
Next up, will GISAID start charging open-source community tools to have access?
That would completely shaft users who contributed to GISAID, where we never agreed to that and assumed GISAID would be good custodians of the data we contributed. They're not.
Next up, will GISAID start charging open-source community tools to have access?
That would completely shaft users who contributed to GISAID, where we never agreed to that and assumed GISAID would be good custodians of the data we contributed. They're not.
October 16, 2025 at 7:48 PM
And now NextStrain as well.
Next up, will GISAID start charging open-source community tools to have access?
That would completely shaft users who contributed to GISAID, where we never agreed to that and assumed GISAID would be good custodians of the data we contributed. They're not.
Next up, will GISAID start charging open-source community tools to have access?
That would completely shaft users who contributed to GISAID, where we never agreed to that and assumed GISAID would be good custodians of the data we contributed. They're not.
Reposted by Robertson Lab
We are at our future planning meeting with Genomics team at main campus of University 🏰 @cvrbioinfo.bsky.social
October 3, 2025 at 4:41 PM
We are at our future planning meeting with Genomics team at main campus of University 🏰 @cvrbioinfo.bsky.social
Reposted by Robertson Lab
Chatbots are explicitly being designed “to elicit intimacy and emotional engagement in order to increase our trust in and dependency on them” (me in Wired magazine 🤭). This designed intimacy combined with AI sycophancy creates serious risks for delusional thinking
www.wired.com/story/ai-psy...
www.wired.com/story/ai-psy...
AI Psychosis Is Rarely Psychosis at All
A wave of AI users presenting in states of psychological distress gave birth to an unofficial diagnostic label. Experts say it’s neither accurate nor needed, but concede that it’s likely to stay.
www.wired.com
September 19, 2025 at 12:45 PM
Chatbots are explicitly being designed “to elicit intimacy and emotional engagement in order to increase our trust in and dependency on them” (me in Wired magazine 🤭). This designed intimacy combined with AI sycophancy creates serious risks for delusional thinking
www.wired.com/story/ai-psy...
www.wired.com/story/ai-psy...
Reposted by Robertson Lab
Very nice reporting on the importance—and the vulnerability— of model organism databases
www.nature.com/articles/d41...
www.nature.com/articles/d41...
Model organism databases face budget cuts and closures
Beyond the crucial data they contain, these digital archives have provided an important space for academic communities to exchange ideas and resources.
www.nature.com
October 2, 2025 at 4:15 PM
Very nice reporting on the importance—and the vulnerability— of model organism databases
www.nature.com/articles/d41...
www.nature.com/articles/d41...
Reposted by Robertson Lab
New exciting resource published!! The Viro3D paper is out, describing our comprehensive database of predicted virus protein structures! 💻🧬
Work with @ulad-litvin.bsky.social , @grovearmada.bsky.social , Alex Jack, @bljog.bsky.social , @davidlrobertson.bsky.social
www.embopress.org/doi/full/10....
Work with @ulad-litvin.bsky.social , @grovearmada.bsky.social , Alex Jack, @bljog.bsky.social , @davidlrobertson.bsky.social
www.embopress.org/doi/full/10....
Viro3D: a comprehensive database of virus protein structure predictions | Molecular Systems Biology
imageimageViro3D provides proteome-level, high confidence AI-protein structure predictions for
>4,400 viruses, allowing mapping of form and function across the human and animal
virosphere. Viro3D i...
www.embopress.org
September 28, 2025 at 5:39 AM
New exciting resource published!! The Viro3D paper is out, describing our comprehensive database of predicted virus protein structures! 💻🧬
Work with @ulad-litvin.bsky.social , @grovearmada.bsky.social , Alex Jack, @bljog.bsky.social , @davidlrobertson.bsky.social
www.embopress.org/doi/full/10....
Work with @ulad-litvin.bsky.social , @grovearmada.bsky.social , Alex Jack, @bljog.bsky.social , @davidlrobertson.bsky.social
www.embopress.org/doi/full/10....
Reposted by Robertson Lab
🚨 New Web Resource Alert! 🚨 We're delighted to share Viro3D a database of >85000 viral protein structure predictions from >4400 human & animal viruses.
🔗 viro3d.cvr.gla.ac.uk
📄 www.embopress.org/doi/full/10....
@molsystbiol.org @cvrinfo.bsky.social @uofgmvls.bsky.social #Virology #AlphaFold 🧪 🦠
🔗 viro3d.cvr.gla.ac.uk
📄 www.embopress.org/doi/full/10....
@molsystbiol.org @cvrinfo.bsky.social @uofgmvls.bsky.social #Virology #AlphaFold 🧪 🦠
Viro3D
viro3d.cvr.gla.ac.uk
September 26, 2025 at 11:36 AM
🚨 New Web Resource Alert! 🚨 We're delighted to share Viro3D a database of >85000 viral protein structure predictions from >4400 human & animal viruses.
🔗 viro3d.cvr.gla.ac.uk
📄 www.embopress.org/doi/full/10....
@molsystbiol.org @cvrinfo.bsky.social @uofgmvls.bsky.social #Virology #AlphaFold 🧪 🦠
🔗 viro3d.cvr.gla.ac.uk
📄 www.embopress.org/doi/full/10....
@molsystbiol.org @cvrinfo.bsky.social @uofgmvls.bsky.social #Virology #AlphaFold 🧪 🦠
Reposted by Robertson Lab
🌊 Coronaviruses likely acquired their spike proteins from aquatic herpesviruses. 4/5
September 26, 2025 at 2:06 PM
🌊 Coronaviruses likely acquired their spike proteins from aquatic herpesviruses. 4/5
Reposted by Robertson Lab
🔍 Searches against structural network are extremely sensitive and allow to recover more RdRps than iterative profile-based searches. 3/5
September 26, 2025 at 2:06 PM
🔍 Searches against structural network are extremely sensitive and allow to recover more RdRps than iterative profile-based searches. 3/5
Reposted by Robertson Lab
🦠 The diversity of viral proteins can be reduced to 19,000 structural clusters. A structure-similarity network captures relationships between them. 2/5
September 26, 2025 at 2:06 PM
🦠 The diversity of viral proteins can be reduced to 19,000 structural clusters. A structure-similarity network captures relationships between them. 2/5