Metagenomics is greatly improving our ability to discover new viruses, as well as their possible associations with disease. However, metagenomics has also changed our understanding of viruses in general. The vast expansion of currently known viral diversity has revealed a large fraction of non-pathogenic viruses, and offers a new perspective in which viruses function as important components of many ecosystems. In this vein, studies of the human blood virome are often motivated by the search for new viral diseases, especially those associated with blood transfusions. However, these studies have revealed the common presence of apparently non-pathogenic viruses in blood, particularly human anelloviruses and, to a lower extent, human pegiviruses (HPgV). To shed light on the diversity of the human blood virome, we subjected pooled plasma samples from 587 healthy donors in Spain to a viral enrichment protocol, followed by massive parallel sequencing. This showed that anelloviruses were clearly the major component of the blood virome and showed remarkable diversity. In total, we assembled 332 complete or near-complete anellovirus genomes, 50 of which could be considered new species. HPgV was much less frequent, but we, nevertheless, recovered 17 different isolates that we subsequently used for characterizing the diversity of this virus. In-depth investigation of the human blood virome should help to elucidate the ecology of these viruses, and to unveil potentially associated diseases.

Anelloviruses are the most common human infectious viruses and are considered essentially harmless. Compared to other human viruses, they are characterized by enormous diversity, and recombination is suggested to play an important role in their diversification and evolution.

The generation of different types of defective viral genomes (DVG) is an unavoidable consequence of the error-prone replication of RNA viruses. In recent years, a particular class of DVGs, those containing long deletions or genome rearrangements, has gain interest due to their potential therapeutic and biotechnological applications. Identifying such DVGs in high-throughput sequencing (HTS) data has become an interesting computational problem. Several algorithms have been proposed to accomplish this goal, though all incur false positives, a problem of practical interest if such DVGs have to be synthetized and tested in the laboratory. We present a metasearch tool, DVGfinder, that wraps the two most commonly used DVG search algorithms in a single workflow for the identification of the DVGs in HTS data. DVGfinder processes the results of ViReMa-a and DI-tector and uses a gradient boosting classifier machine learning algorithm to reduce the number of false-positive events. The program also generates output files in user-friendly HTML format, which can help users to explore the DVGs identified in the sample. We evaluated the performance of DVGfinder compared to the two search algorithms used separately and found that it slightly improves sensitivities for low-coverage synthetic HTS data and DI-tector precision for high-coverage samples. The metasearch program also showed higher sensitivity on a real sample for which a set of copy-backs were previously validated.