Expand genome metadata in Ensembl with AI tools

The Ensembl Plants and Metazoa platforms face a significant metadata gap where critical biological context, such as ploidy, strain, and sex, is often documented in peer-reviewed literature but missing from formal INSDC sequence archives. This lack of structured metadata, particularly ploidy, creates technical bottlenecks in downstream comparative genomics and requires labor-intensive manual curation. To address this, I propose developing a standalone, Python-based "Retrieve-Extract-Verify" module that leverages Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) to automate metadata extraction from scientific papers. This tool will utilize literature APIs like Europe PMC, employing section-aware parsing and confidence scoring to ensure high data integrity with minimal human intervention. My previous experience in building Transformer-based bioprocess optimization models and genome-level embeddings directly informs my approach to handling complex biological datasets. The final outcome will be a ready-to-deploy Nextflow module designed for seamless integration into Ensembl’s production pipeline and provide the global research community with richer, more reliable genomic metadata.