I-TASSER

I-TASSER (Iterative Threading ASSEmbly Refinement) is a unified platform for automated protein structure and function prediction. It utilizes a hierarchical protocol that begins with fold recognition via LOMETS (threading), followed by fragment-based assembly through replica-exchange Monte Carlo simulations. The system is particularly noted for its ability to bridge the gap between template-based modeling and ab initio folding by refining structural decoys into high-resolution models. In the 2026 landscape, I-TASSER has integrated deep-learning spatial constraints (D-I-TASSER) and multi-domain threading capabilities (I-TASSER-MTD) to compete with end-to-end transformers like AlphaFold. Its architecture is uniquely suited for functional annotation, utilizing the COACH algorithm to predict ligand-binding sites and Gene Ontology (GO) terms by matching the generated 3D models against the BioLiP database. It remains a critical tool for researchers working with proteins that lack clear homologs, providing structural insights where pure sequence-based methods often falter. The platform operates primarily through a high-performance computing cluster at the University of Michigan, offering both a web-based interface and standalone local installations for enterprise-scale genomic pipelines.