Overview
Robetta is a protein structure prediction service leveraging both deep learning and comparative modeling techniques. At its core is RoseTTAFold, a deep learning-based method renowned for its speed and accuracy in predicting protein structures from sequence. The service offers an interactive submission interface that accommodates custom sequence alignments, homology modeling constraints, and local fragments. Robetta can model multi-chain complexes using either RoseTTAFold (requiring paired MSA) or comparative modeling (CM) approaches. The CM method utilizes the PDB100 template database, a co-evolution based model database (MDB), and allows for custom templates. Computing resources are provided by the Baker lab and Rosetta@home, a distributed computing project. The service facilitates large-scale sampling, making it valuable for research into protein function, drug discovery, and understanding biological mechanisms. Its architecture is designed to incorporate community contributions and leverage volunteer computing power.