AI-Assisted Curation Assistant Tool for cBioPortal

This project aims to extend the existing cBioAbstractor prototype into a schema-driven formatting pipeline that assists curators in transforming heterogeneous supplemental datasets into cBioPortal-compatible formats. Currently, curation is a manual and time-consuming process due to inconsistent file structures, lack of automated schema mapping, and the need to interpret validation errors manually. This results in inefficiencies, variability, and high onboarding effort. My preliminary extension implementation suggests the feasibility of this approach by connecting existing components, introducing fuzzy column normalization, and providing a simple interface for file upload, transformation, and result inspection. Initial qualitative experiments show that the system can reconstruct correct cBioPortal formats from intentionally perturbed datasets, suggesting strong potential for reducing manual work and improving consistency.