Optimization of viral reference databases for NGS-based viral safety testing reduces background signals while preserving detection sensitivity.

Abstract

Next-generation sequencing (NGS) enables comprehensive analysis of nucleic acids present in a sample and is considered a promising tool for viral safety testing of biologics. For this purpose, reference viral databases are required to provide broad coverage of potential contaminants. However, in our previous study, mapping RNA sequencing data from adenovirus-infected cells to a highly comprehensive viral database resulted in the detection of the spiked adenovirus as well as numerous additional viral sequences (background viruses). In this study, we constructed a curated viral reference database by restricting the included sequences primarily to viral genomes available in the RefSeq database. We evaluated its performance in terms of detection sensitivity for the target virus represented in the database, the number of background viruses, and the ability to detect viruses that were not included in the database. Our results demonstrate that optimizing reference database composition can significantly reduce background signals while preserving detection capability, highlighting the importance of database design in NGS-based viral safety testing.