Evaluation of short-read-only, long-read-only, and hybrid assembly approaches on metagenomic samples demonstrating how they affect gene and protein prediction which is relevant for downstream functional analyses. For a human gut microbiome sample, we use complementary metatranscriptomic, and metaproteomic data to evaluate the metagenomic-based protein predictions.