Higher-Order Protein Feature Co-occurrence at AlphaFold Scale

Preprint: ET-miner discovers which combinations of protein features (Pfam domains, Gene Ontology terms, structural properties) co-occur across the AlphaFold protein universe. Using GPU-resident frequent itemset mining, ET-miner exhaustively mines 76.9 million multi-feature proteins in ~44 minutes on a single NVIDIA RTX3090, revealing 26.8 million co-occurrence patterns up to K=22 items deep. The deepest pattern describes a neuronal antiviral RNA helicase sentinel shared by exactly 8 proteins. This work demonstrates that exact Apriori computation at 100-million-transaction scale is practical on modern consumer-hardware GPUs.