Dan Liu
@danliu1.bsky.social
Computational biologist | Bioinformatics, virus-host interactions, LLMs 🦠 💻
Thanks to the fantastic AI-in-bio community at the @cvrinfo.bsky.social, @uofgcancersciences.bsky.social
@uofgterrierteam.bsky.social
@uofgterrierteam.bsky.social
October 28, 2025 at 3:27 PM
Thanks to the fantastic AI-in-bio community at the @cvrinfo.bsky.social, @uofgcancersciences.bsky.social
@uofgterrierteam.bsky.social
@uofgterrierteam.bsky.social
A huge thanks to Craig Macdonald, @davidlrobertson.bsky.social and Ke Yuan for supervising this work, and other co-authors — Fran Young, @kieranlamb.bsky.social, @adalbertocq.bsky.social, Alexandrina Pancheva, and Crispin Miller.
October 28, 2025 at 3:27 PM
A huge thanks to Craig Macdonald, @davidlrobertson.bsky.social and Ke Yuan for supervising this work, and other co-authors — Fran Young, @kieranlamb.bsky.social, @adalbertocq.bsky.social, Alexandrina Pancheva, and Crispin Miller.
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.
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.
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.
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.
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.
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.
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.
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).
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 12:32 AM
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.
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 12:32 AM
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.
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 12:32 AM
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.
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 12:32 AM
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.
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 12:32 AM
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.
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 12:32 AM
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.
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 12:32 AM
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.
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 12:32 AM
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).
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 12:24 AM
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.