LoGoPPI enables fast and accurate protein protein interaction mapping at scale
Lee, H. B.; Ma, J.; Kim, H.-J.; Song, K.; Lee, I.
Show abstract
Graph-based protein function analysis is powerful, but protein-protein interaction (PPI) networks exist for only a small fraction of animal and plant genomes. We present LoGoPPI, which infers PPIs from sequence by combining bi-encoder global protein representation with local residue-level late interaction. LoGoPPI matches or exceeds state-of-the-art PLM-based cross-encoders while achieving orders-of-magnitude faster inference, up to ~1,500-fold, and its local branch provides residue-level signals associated with interaction interfaces and structurally flexible regions. This efficiency enables practical proteome-wide PPI reconstruction at scales prohibitive for cross-encoder models, potentially extending interactome mapping to tens of thousands of animal and plant species. LoGoPPI thus provides a scalable framework for comparative and functional analysis of protein networks across diverse taxa.
Matching journals
The top 3 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Learning multi-cellular representations of single-cell transcriptomics data enables characterization of patient-level disease states 94%
- Inferring high-dimensional pathways of trait acquisition in evolution and disease 94%
- Identifying maximally informative signal-aware representations of single-cell data using the Information Bottleneck 93%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.