BeeNA: um pipeline automatizado para anotação e identificação de vírus de RNA

Abstract

Studies on RNA viruses in various environments using RNA sequencing have been increasing, but the true diversity of the virosphere remains poorly understood. With the growing availability of public data in databases originated from metagenomics sequecing, new viruses have been discovered in both well- known and less explored organisms. Currently, the main pipelines for RNA virus identification and annotation, VirBot and VirID, face limitations due to their use of Hidden Markov Model (HMM) profiles and sequence similarity analysis, which result in false negatives. This study proposes an automated pipeline based on the Snakemake platform, combining the high-performance viral identification tool geNomad with the DIAMOND method for sequence similarity analysis. In tests with artificial samples, the pipeline achieved >80% sensitivity for species- level identification and >90% for family- and genus-level identifications. Compared to other strategies, VirBot was unable to identify any sample at the species level, achieving 64% sensitivity at the family level. BeeNA showed better performance with the Megahit and rnaSPAdes assemblers, while tools like coronaSPAdes were ineffective. The pipeline also allows metagenomic analyses of non-viral organisms, viral quantification, and automated confirmation of potential “new” species through phylogenetic analysis, despite limitations in identifying highly divergent viruses due to the use of Pfam pHMMs. Benchmarking with real metagenomic data allowed for the detection of organisms previously found and not found in original studies on microorganisms from bees and humans, demonstrating the pipeline’s potential for viral identification. However, to enhance species-level identification and RdRp detection for phylogenetic analysis, it is essential to calibrate geNomad and utilize more extensive HMM datasets, such as RVDB-prot.