BioBlobs: Unsupervised Discovery of Functional Substructures for Protein Function Prediction

Protein function is driven by cohesive substructures, such as catalytic triads, binding pockets, and structural motifs, that occupy only a small fraction of a protein's residues. Yet existing pipelines built on protein encoders do not model proteins at the substructure level, leaving the central biological question unanswered: \emph{which substructure of a protein is responsible for its function?} We introduce \tool, an encoder-agnostic, end-to-end differentiable framework that compresses a protein into a small set of cohesive substructures (\emph{blobs}) and predicts function from these blobs alone, so that each blob corresponds to a candidate functional region. Across diverse protein function prediction tasks and multiple sequence- and structure-based encoders, \tool matches or exceeds strong baselines while operating on only a small fraction of residues. The discovered \emph{blobs} adapt their spatial scale to the task, ranging from local catalytic sites to entire structural domains. Trained only on protein-level labels, \tool recovers experimentally annotated catalytic sites in the M-CSA database, demonstrating unsupervised functional substructure discovery and opening a path to large-scale functional site discovery across the unannotated proteome.