During the investigation of potential pathogenic viruses in humans, three usable libraries were downloaded from the SRA database (SRR16633776, SRR16633740, SRR16609930), submitted by Cordey et al. (2021) from the University Hospitals of Geneva. The sample processing methods have been described in previous publications (Cordey et al., 2021). SRA format files were converted to fastq format using Pfastq-dump v0.1.6 (https://github.com/inutano/pfastq-dump), and host sequences were removed using Bowtie2 v2.4.5 (Langmead and Salzberg, 2012; Langmead et al., 2019). Additionally, potential primer sequences in the raw reads were trimmed using Trim Galore v0.6.5 (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore) and the subsequent files underwent quality control assessments employing parameter “–phred33 –length35 –stringency3 –fastqc.” Duplicate reads were marked using PRINSEQ-lite v0.20.4 (-derep 1) (Schmieder and Edwards, 2011), and the final library was assembled by an internal pipeline. Using default parameters, single-end reads were assembled with MEGAHIT v1.2.9 (Li et al., 2016). The resulting assemblies were subsequently imported into Geneious Prime v2022.0.1 for classification and manual confirmation (Kearse et al., 2012). Individual overlapping clusters serve as references for mapping to the original data using low sensitivity/fastest parameters in Geneious Prime. Additionally, mixed assembly was performed using MEGAHIT in conjunction with BWA v0.7.17 to search for potential low-abundance overlapping clusters from the unused reads (Buchfink et al., 2015).

The blastx program within DIAMOND was utilized to compare alleles against the non-redundant protein database (nr). Based on the blastx program established in DIAMOND v2.0.15 (Buchfink et al., 2021), an E-value threshold of < 10⁻⁵ was applied. Additionally, protein sequences such as RdRp (RNA-dependent RNA polymerase), Rep (replication-associated protein), and NS1 (non-structural protein) were downloaded from the RefSeq database. These sequences were used to compare overlapping groups with lengths >1,500 bp. Taxonomic identification was performed using the rma2info program embedded in MEGAN6 (Gautam et al., 2022). The predicted open reading frames (ORFs) were forecasted using Geneious Prime (Kearse et al., 2012) and were compared with relevant viruses. The annotation of these ORFs relied on comparisons with the Conserved Domain Database (CDD). Using MEGAN6 (Huson et al., 2007), a standardization and comparison of the composition analyses for three libraries was conducted, and the results on viral community structure were described using the R v4.2.1 packages pheatmap.

To infer phylogenetic relationships, nucleotide and protein sequences of reference strains from various hosts belonging to the respective viruses were downloaded from the NCBI GenBank database. Relevant nucleotide and protein sequences were aligned using the alignment program implemented in CLC Genomics 10.0 (https://digitalinsights.qiagen.com), and further optimized using MUSCLE in MEGA-X v10.1.8 (Kumar et al., 2018) and the E-INS-I algorithm in MAFFT v7.3.1 (Kuraku et al., 2013). Protein sequences were used to construct a Bayesian inference tree with MrBayes v3.2 (Ronquist et al., 2012), running the Markov chain for up to one million generations with sampling every 50 generations, discarding the first 25% of the Markov Chain Monte Carlo (MCMC) samples as burn-in. A maximum likelihood tree was also constructed to confirm all Bayesian inference trees using MEGA-X v10.1.8 (Kumar et al., 2018). The nucleotide sequences were analyzed using Sequence Alignment Tool v1.2 (Muhire et al., 2014) to determine the color-coded distance matrix between the three Cyclovirus sequences we excavated and other members of the Cyclovirus genus.

ColabFold (Mirdita et al., 2022) was utilized to predict the three-dimensional structures of the viral structural proteins identified in this study, while UCSF ChimeraX (Goddard et al., 2018) was employed for visualization and analysis.

The three libraries generated a total of 30,006,599 raw reads on the Illumina HiSeq platform, resulting in 29,999,462 clean reads after quality control. A comparison of the clean reads with the nr database revealed 11,708,463 reads that matched the best viral protein, constituting 39.03% of the total clean reads. Ten viral families were detected, with the most abundant being Anelloviridae (790,177 reads, 93.34%), followed by Siphoviridae (148,280 reads, 1.75%), Myoviridae (125,450 reads, 1.48%), Herpesviridae (12,474 reads, 1.47%), Podoviridae (9,091 reads, 1.07%), Circoviridae (4,358 reads, 0.50%), Phycodnaviridae (2,064 reads, 0.24%), Mimiviridae (601 reads, 0.07%), Ascoviridae (340 reads, 0.04%), and Autographiviridae (223 reads, 0.02%). Additionally, there were 428,994 reads that could not be assigned to any taxonomic level (E-value > 10⁻⁵), which may represent potential novel viruses. Heatmap illustrates the differences in abundance among the three groups (SRR16609930, SRR16633740, and SRR16633776) across various viral families. The figure indicates that the abundance of Anelloviridae was consistently high in all three groups, exceeding 5, particularly in the SRR16609930 group, where it reached 5.47. Notably, Anelloviridae has been linked to immune modulation and inflammation, with higher viral loads observed in individuals with autoimmune conditions or chronic infections, such as HIV and hepatitis (Focosi et al., 2024). This could explain the observed high abundance across the groups, as immune status and chronic conditions are often correlated with elevated Anelloviridae levels. In contrast, the abundance of Herpesviridae was moderate in SRR16609930 and SRR16633740, but significantly increased in SRR16609930 to 4.00 (Figure 1). Previous research has demonstrated that Herpesviridae levels can fluctuate with host immune status, particularly during periods of immune suppression or reactivation. This suggests that the higher abundance of Herpesviridae in SRR16609930 may be related to specific immune conditions or viral reactivation in the host. Similarly, Myoviridae and Podoviridae exhibited inter-group variations, with SRR16633776 showing higher abundance of Myoviridae compared to the other groups. Furthermore, certain viral families, such as Ascoviridae, were undetected in SRR16609930 but were found in the SRR16633776 group at an abundance of 2.53. This observed variability in viral family abundance underscores the complex interplay between host immune factors and environmental conditions, and suggests that both biological and ecological factors contribute to the presence and diversity of viral families in these groups.

In this study, the sequences of Circoviridae (4,358 reads) were subjected to de novo assembly using Geneious Prime v2019.0.4, successfully yielding three complete genomes associated with the family Circoviridae, designated as SRR16609930_k99_603 (1,873 bp), SRR16633740_k79_101 (1,728 bp), and SRR16633776_k99_746 (1,725 bp). Each of these complete genomes contains two open reading frames (ORFs), which encode the Rep and Capsid proteins (Rosario et al., 2017). The accuracy of these assemblies was validated through BLASTx comparison (Figure 2). This finding provides significant genomic data for the classification and functional studies of Circoviridae.

According to the distance matrix data, there is a high similarity between SRR16633740_k79_101 and SRR16633776_k99, both exhibiting the highest similarity score of 97.05% with NC_032682 (Figure 3). Additionally, the similarity score between AB937980 and SRR16609930_k99_603 is 92.52%. These results indicate a relatively high similarity between the Circoviridae viral sequences of Mastomys natalensis and those of human Circoviridae. Notably, the similarity scores for the Rep protein gene sequences of other known Circoviridae viruses from the African region are all above 90% when compared to SRR16609930_k99_603. These data suggest that this viral group may primarily circulate between animals and humans in Africa.

In this study, a phylogenetic tree was constructed based on the Rep amino acid sequences of the newly discovered cycloviruses to elucidate their relationships with known viral families (Varsani et al., 2024). By comparing the Rep sequences of these newly identified cycloviruses with sequences from established families, we were able to infer their evolutionary relationships and position within the current viral taxonomy. Coupled with BLASTx results from the NCBI database, we successfully classified these three newly recognized complete cycloviruses under the genus Cyclovirus within the Circoviridae family (Figure 4).

Notably, phylogenetic analysis revealed that SRR16633740_k79_101 and SRR16633776_k99_746 form a distinct cluster with the known NC_032682. The BLASTx comparison indicated similarity scores of 93.24% and 93.67%, respectively, with NC_032682, derived from cerebrospinal fluid in India, while similarity to other known viral family sequences was below 65%, suggesting the potential formation of a new subfamily. Additionally, SRR16609930_k99_603 displayed a similarity of 91.77% with AB937980, obtained from the feces of Mastomys natalensis in Zambia, and both showed ≥87% similarity to various viral sequences from African animals and humans. These data support the hypothesis that this viral group predominantly circulates between animals and humans in Africa, with the phylogenetic clustering of these newly discovered cycloviruses in close relation to known African viral species suggesting potential zoonotic spillover.

The phylogenetic distances observed between these cycloviruses and other related viral sequences suggest that while these new viruses share a recent common ancestor with known species, their distinct clustering also points to a possible divergence driven by ecological and evolutionary pressures, such as host adaptation and environmental factors. The high similarity to viral sequences found in both humans and animals within Africa indicates that the virus may have a broad host range, and the observed phylogenetic patterns may reflect ongoing interspecies transmission. This further supports the idea that these cycloviruses could be circulating between human and animal populations, potentially through direct contact or shared ecological niches, and may be subject to selective pressures that favor their persistence across diverse host species.

These findings underscore the zoonotic potential of these cycloviruses and the need to investigate potential transmission routes, such as spillover events from animal to human populations. The clustering of viral sequences from Africa also raises questions about the role of regional ecological factors—such as wildlife trade, habitat encroachment, or changes in animal migration patterns—in facilitating cross-species transmission. Further studies examining the host range, transmission dynamics, and evolutionary pressures acting on these cycloviruses are essential to understanding their ecological and public health significance.

The Rep protein of the Circoviridae family plays a crucial role in the viral lifecycle, with primary functions including promoting the replication of viral DNA as a vital enzyme, regulating viral RNA transcription to enhance protein synthesis, participating in RNA splicing to generate mature mRNA, and influencing host cell metabolism and immune responses (Luo et al., 2018). To predict and compare the spatial structures of the Rep proteins identified in this study with known sequences, we downloaded NC_032682 and AB937980 from the GenBank database and utilized the ColabFold tool for spatial structure predictions (Figure 5).

The structural similarity between SRR16633740_k79_101, SRR16633776_k99_746, and NC_032682 is striking, indicating not only evolutionary conservation but also functional similarity. The overall conformations of these Rep proteins are highly similar, with well-aligned α-helices and β-sheets, suggesting that these proteins likely perform similar roles in viral replication and host interaction. This conservation points to the stability of the structural elements critical for DNA replication, as well as for the modulation of host immune responses. Given the critical role of Rep in these processes, such structural consistency is likely necessary for maintaining viral fitness and the virus's ability to evade immune detection and modulate host cell activities.

However, SRR16609930_k99_603 and AB937980, despite showing a sequence similarity of 91.77%, exhibit notable structural differences. These differences are most evident in the spatial arrangement of certain key α-helices and β-sheets, which appear to be displaced or altered compared to NC_032682. Specifically, mutations in a few critical amino acids likely lead to shifts in these secondary structure elements, resulting in a reorganization of the protein's overall tertiary structure. These changes could have profound implications for the Rep protein's function. For instance, the altered structural configuration may affect the protein's ability to interact with host replication machinery, potentially reducing the efficiency of viral DNA replication. Additionally, structural differences in the Rep protein could also impact the virus's ability to infect different host species. If these mutations alter how the Rep protein binds to host cell receptors or interact with host immune components, it could enable the virus to adapt to new host environments, thus facilitating zoonotic spillover or cross-species transmission.