Back

PRIorI: a graph-based mining of structural arrangements in protein-protein interfaces

Ribeiro, V. S.; Paiva, V. A.; Gomes, I. S.; Santos, L. M.; Silva, H. H. M.; Dias, V. V. S.; Silveira, S. A.

2025-12-09 bioinformatics
10.64898/2025.12.03.692053 bioRxiv
Show abstract

SummaryProtein-protein interactions (PPIs) are fundamental to biological processes and central to understanding disease mechanisms, making them essential for drug discovery, peptide-based therapeutics, and vaccine development. Identifying conserved structural arrangements within PPI interfaces can provide valuable insights into molecular recognition and interaction mechanisms. Here, we introduce PRIorI (PRotein-PRotein InteractiOn gRaph Isomorphism), a web-based platform to explore precomputed protein-protein interaction networks and identify conserved atomic-level interaction motifs within structural complexes. Unlike traditional methods, PRIorI models PPI interfaces as bipartite graphs, applying graph isomorphism techniques to efficiently retrieve interaction patterns independent of sequence alignment or structural superimposition. Users can query precomputed PDB interaction graphs, upload custom protein structures, or design structural motifs for targeted searches. To illustrate its applicability, we used PRIorI to analyze the SARS-CoV-2 Spike-ACE2 interface, identifying key interaction motifs previously reported in the literature. The platform successfully identified salt bridges, hydrogen bonds, and hydrophobic interactions that stabilize the viral-host complex, demonstrating PRIorIs utility in protein interaction analysis. Availability and implementationWebserver implemented using typescript, python, Next.js and Express.js. Uses MongoDB for storage, Redis for cache and is hosted by Apache on an Ubuntu Server. PRIorI web platform is freely available at https://priori.ufv.br/. Contactsabrina@ufv.br Supplementary informationSupplementary data are available at Bioinformatics online.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"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.