Dog-wise canine gut metagenome assemblies with reconstructed bacterial genomes and viral candidates
TL;DR
DogMAG is described, a canine gut metagenome resource based on dog-wise long-read and hybrid assemblies generated by grouping sequencing libraries according to canonical dog identity before assembly that provides assemblies, genome and viral candidate sequences, metadata, provenance tables and workflow scripts for reuse, benchmarking and reanalysis.
Abstract
Long-read metagenomic sequencing can improve genome recovery from complex gut microbial communities, yet directly reusable canine gut genome resources remain limited. Here we describe DogMAG, a canine gut metagenome resource based on dog-wise long-read and hybrid assemblies generated by grouping sequencing libraries according to canonical dog identity before assembly. The final dataset comprises 41 assemblies linked to 277 FASTQ records, including 30 Flye long-read-only and 11 OPERA-MS hybrid assemblies. A single integrated BASALT workflow produced 11,276 selected bin/version records, followed by explicit quality-based re-selection of 3,418 medium-quality-or-better metagenome-assembled genome candidates. External dRep dereplication yielded 792 strain-like representatives at 99% average nucleotide identity and 135 species/SGB-like representatives at 95%. GTDB-Tk classified all 792 representatives as Bacteria. Viral screening identified 22,068 geNomad predictions, of which 3,374 Complete, High-quality or Medium-quality viral/proviral candidate rows passed CheckV filtering with contamination ≤10%. DogMAG provides assemblies, genome and viral candidate sequences, metadata, provenance tables and workflow scripts for reuse, benchmarking and reanalysis.