SUMMARY RNA-protein interactions mediate multiple critical regulatory steps in the translation of information from the genome to the cellular machine, and corruptions of these fundamental interactions are implicated throughout infectious and inherited disease, neurodegeneration, and cancer. Unfortunately, a scarcity of predictive computational tools for RNA-protein interactions is slowing the development of potentially life-saving efforts that either target or repurpose these interactions to address human disease. This proposal brings together four labs to resolve this bottleneck, building on our recent studies that have achieved – all for the first time – blind protein-RNA structure predictions reaching near-atomic resolution, large-scale prediction of protein-RNA binding energetics with accuracy and precision of better than 1 kcal/mol, and redesign of a complex protein- RNA interface to accurately retarget silencing complexes in vivo. We propose herein to unify, rigorously test, and disseminate our labs’ methods to tackle three separate but synergistic computational problems in protein- RNA research: automated correction of errors that pervade experimental protein-RNA complex structures (Aim 1), our richest resources of protein-RNA information; prediction of impacts of mutation on RNA-protein interaction energetics (Aim 2) that could highlight new regulatory links in disease-associated alleles; and design of novel engineered RNA-protein interactions (Aim 3) to facilitate rational perturbation of genetic events to aid biological inquiry and eventually to ameliorate disease. We will evaluate success within each of our Aims through true blind predictions tested through rapidly emerging cryoelectron microscopy maps and repurposed sequencers that can measure hundreds of thousands of RNA-protein affinities in single experiments and, more broadly, by adoption of our Rosetta software and online tools by the general biomedical research community. The proposed protein-RNA-focused research addresses an area of molecular modeling that has received surprisingly little attention in the computational community but is unambiguously important for accelerating biological understanding and molecular medicine.