This document discusses using text mining to identify cancer-gene relationships from biomedical literature to help precision cancer medicine. It describes a pipeline to extract these relationships from PubMed abstracts using named entity recognition and relationship extraction. The extracted relationships are then prioritized to suggest new items to add to CIViC, a community curated knowledge base. Evaluation found the prioritized lists contained multiple items that should be added to CIViC. The goal is to continually update the extracted relationships as new literature is added to PubMed.