BGC-XPLORER: Exploring Biosynthetic Gene Clusters with Gene-Level Functional Annotation and AI-Assisted Summarization

Turning one newly sequenced bacterial genome into a view of its biosynthetic potential still means running several cluster detectors and annotation tools and comparing their output by hand. We present BGC-XPLORER, a containerized Snakemake workflow for a single assembled bacterial genome, deployed locally with Docker and operated through a browser. It annotates the genome with Bakta, runs antiSMASH, GECCO and DeepBGC, and groups their predictions gene-by-gene, defining a locus by the genes that at least two callers agree on. Each locus carries gene-resolved evidence from Bakta, eggNOG and dbCAN, ARTS self-resistance hits and MIBiG comparisons in an interactive report with searchable tables and annotated gene maps. Against 42 MIBiG reference loci in four actinomycete genomes from two genera, the grouping recovered 41, at least as many as any single caller in every genome. Applied to the newly sequenced soil isolate Streptomyces albidoflavus STR-S001 (GenBank CP158056), the workflow reported 49 loci, of which 27 had support from at least two callers, 22 a MIBiG comparison and five an ARTS known-resistance hit. An optional, remotely generated AI summary of a single locus is offered as an aid to reading the report. The workflow is not a replacement for laboratory verification but a complementary step that precedes it.