Benchmark Datasets for GCVR: Confidence-Aware Viral Identification through Graph Propagation and Functional Priors

Benchmark datasets used to evaluate GCVR, a graph-based framework for viral sequence identification from metagenomic contigs. This record contains the RefSeq-Artificial and NCBI-SimAssembly datasets, including the final contig sequences and corresponding labels used in the experiments. It also includes the PHROG functional-prior weights used by GCVR. The RefSeq-Artificial dataset was derived from publicly available sequence resources, whereas the NCBI-SimAssembly dataset was generated in this study from NCBI reference genomes using a simulated short-read metagenomic sequencing and assembly workflow. The complete CAMI2-Marine dataset used in this study is provided in the GCVR GitHub repository.