<p>Targeted genome mining to expand known families of natural products is a powerful strategy for discovering bioactive compounds, yet it remains a significant bioinformatics challenge. While tools exist for de novo biosynthetic gene cluster identification and large-scale unsupervised clustering, dedicated methods for the targeted, hypothesis-driven expansion of user-defined BGC families are lacking. Here, we present DiscERN (Discoverer of Evolutionarily Related Natural products), a user-friendly tool designed to address this gap. DiscERN leverages a multi-modal ensemble method that integrates four complementary algorithms classifying biosynthetic gene clusters based on Pfam content, sequence homology, and predicted product structure. This approach allows users to strategically balance discovery sensitivity with predictive precision to suit diverse research goals. We demonstrate DiscERN’s utility by applying it to a large collection of actinomycete genomes and validating its predictive power through the successful isolation of discomycin A, a new calcium-dependent lipopeptide antibiotic, from a silent biosynthetic gene cluster. DiscERN provides a robust and accessible platform that streamlines the path from genomic data to a prioritised list of candidate biosynthetic gene clusters, effectively bridging the gap between in silico prediction and bioactive compound discovery.</p>