Detecting RNA Viruses in Cattle: Effects of Sequencing Depth and Sequence References in Metatranscriptomics

Detecting RNA Viruses in Cattle: Effects of Sequencing Depth and Sequence References in Metatranscriptomics | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Detecting RNA Viruses in Cattle: Effects of Sequencing Depth and Sequence References in Metatranscriptomics Barbara Brito, Melinda Frost, John Webster, Joyce To, Peter Kirkland This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7071637/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract In this study, the impact of sequencing depth and reference genome selection is shown to influence the detection of five bovine respiratory RNA viruses in metatranscriptomes from clinical cattle samples. Metatranscriptomes were generated from pooled samples and subsampled to 20 million, 10 million, and 1 million reads. We assessed the correlation between qRT-PCR cycle threshold (Ct) values and the number of reads mapped to reference genomes for four viruses: Bovine coronavirus (BCoV), Bovine nidovirus (BNV), Influenza D Virus (IDV), and Bovine Viral Diarrhea Virus-1 (BVDV-1). Strong linear correlations were observed between mapped reads and Ct values. For BCoV and BNV, RefSeq genomes yielded detection thresholds at ~ Ct 38 with 1 million reads; IDV showed a threshold at Ct 34.4. Reference genome choice had minimal impact on BCoV, BNV, and IDV detection. However, BVDV-1 detection was poor using the divergent RefSeq genome and improved significantly with a sample-derived reference. Complete coverage of the BCoV genome and IDV segment 6 was achieved in samples with Ct values below 30, regardless of the reference used. For BNV, genome 80% coverage was reached when using the NCBI RefSeq, even in samples with low-Ct samples. BVDV-1 could not be detected when using the RefSeq genome, highlighting the limitations of distant references. These findings demonstrate that both sequencing depth and reference genome choice substantially influence viral detection sensitivity in metatranscriptomic analyses. metagenomics diagnostics sensitivity qRT-PCR reference viral genome Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Metatranscriptomics, the untargeted sequencing of RNA from clinical samples, is increasingly used for the detection and characterization of RNA viruses. This approach enables the identification of both known and novel viruses without prior knowledge of their genomic sequences, making it particularly useful for outbreak investigations and surveillance. While metatranscriptomics offers a comprehensive approach to viral detection, its diagnostic accuracy compared to conventional methods remains insufficiently characterized. Meta-analyses of untargeted metagenomic diagnostics in human medicine—often evaluating blood, cerebrospinal fluid, orthopedic, and respiratory samples—have reported overall sensitivities ranging from 0.75 to 0.90 and specificities between 0.67 and 0.96, depending on sample type (Govender et al. 2021 ; Liu et al. 2022 ; Neyton et al. 2023 ). However, these evaluations have been limited to human clinical contexts, and comparable studies in veterinary medicine—particularly for RNA viruses in livestock—are largely lacking. Although metagenomic approaches are increasingly being explored for diagnostic purposes, their integration into routine veterinary investigations remains limited. As a result, there is a pressing need for systematic evaluations that benchmark their performance against established diagnostic standards such as quantitative reverse transcription PCR (qRT-PCR). While qRT-PCR remains the gold standard due to its high sensitivity and specificity, it is inherently limited to the detection of known viral targets. In contrast, metatranscriptomics enables untargeted detection of both expected and novel RNA viruses, but its diagnostic accuracy—particularly in relation to viral genome divergence and sequencing yield per sample—has not been rigorously assessed. A key technical consideration is how the quantity of sequencing data (sequencing depth) influences the sensitivity of detection. Although high-throughput sequencing platforms can produce substantial read counts, practical and economic constraints often limit the number of reads per sample. Consequently, there is a need to determine how sequencing depth affects detection sensitivity, particularly across a range of viral loads. Another important consideration in metatranscriptomic analysis is the choice of reference genome. While some bioinformatic pipelines incorporate de novo assembly and taxonomic classification of contigs, these approaches are computationally intensive and more difficult to automate compared to direct mapping to reference genomes (Shakya et al. 2019 ). However, the accuracy of read-mapping–based detection depends heavily on the sequence similarity between the virus present in the sample and the chosen reference genome (Cobbin et al. 2021 ). For many veterinary pathogens—particularly RNA viruses—there is limited representation in public databases, and field strains often differ significantly from available references. The extent to which this genomic divergence impacts detection sensitivity remains poorly characterized. To address these gaps, this study evaluates the diagnostic performance of metatranscriptomics for detecting four clinically relevant bovine respiratory RNA viruses—Bovine Coronavirus (BCoV), Bovine nidovirus (BNV), Influenza D Virus (IDV), and Bovine Viral Diarrhea Virus-1 (BVDV-1). Using pooled nasal and nasopharyngeal samples, we compared mapped read counts against qRT-PCR cycle threshold (Ct) values across subsampled sequencing depths (20M, 10M, and 1M reads). We further assessed how reference genome choice—comparing NCBI RefSeq entries with viral genomes assembled from the same dataset—affects detection and genome coverage. This study aims to provide critical insights for optimizing metagenomic approaches in veterinary diagnostics. Methods Samples and data The study samples included nasopharyngeal and nasal swabs collected in transport medium designed to preserve nucleic acid (Copan® ENAT and Copan® UTM) from cattle in a feedlot in New South Wales, Australia. The libraries used in this study are detailed in Brito et al., 2023 (Brito et al. 2023 ). In brief, pooled RNA extracted from nasal swabs from animals with or without bovine respiratory disease were analysed. RNA extraction was performed using the RNeasy Plus Micro Kit and RNeasy Plus Mini Kit (Qiagen). Libraries were prepared using the TruSeq Stranded Total RNA Human/Mouse/Rat Kit (Illumina) and sequencing was conducted on a NovaSeq S1 (Illumina) platform at the Australian Genome Research Facility. The total read count per sample and corresponding Bioproject details are provided in Table S1. RT-qPCR Four quantitative real-time PCR (qRT-PCR) targeting Bovine nidovirus (BNV), Bovine Coronavirus (BCoV), Influenza D virus (IDV), and Bovine Viral Diarrhea-1 (BVDV) were conducted on non-pooled samples. These samples were pooled before next generation sequencing library preparation as described above. The Ct values for each of the 15 pools was estimated from the combined values of the individual samples that comprised each pool. To estimate the cycle threshold (Ct) values equivalent to the pooled libraries, the total copy number of each contributing sample was calculated using a 95% efficiency slope equation, Ct = − 3.45x + 40. The cumulative copy number from all samples was subsequently converted back into a corresponding Ct value for each pooled library. Impact of sequencing depth and reference sequences in viral detection The raw sequences were initially quality trimmed using BBDuk from BBTools 38.87 (Bushnell 2014 ). To evaluate how the depth and reference sequence selection influence viral detection, each of the QC trimmed metatranscriptomes were randomly subsampled to approximate different sequencing depths. Subsamples of 20M, 10M, and 1M reads were generated from each library using seqtk sample (with the -s100 seed option) (github.com/lh3/seqtk). The subsampled reads were mapped to both, the viruses assembled in this study and the viral species RefSeq from NCBI for Bovine nidovirus (NC_027199.1), Bovine Coronavirus (NC_003045.1), Influenza D virus (NC_036615-21) and Bovine Viral Diarrhea-1 (NC_001461.1) using the BWA-MEM2 aligner (Md et al. 2019 ). A mapping quality threshold of an alignment score > = 100 was applied to retain high-quality mapped contigs and Samtools was used to compute the mapping statistics and read coverage for each reference (Fig. 1 ) (Danecek et al. 2021 ). For the segmented virus Influenza D, mapped reads were obtained for all seven segments to determine if there were major differences in read mapping by segment. Correlation between mapped reads and qRT-PCR, and differences across reference genomes qRT-PCR was performed for the four viruses mentioned above, using a limit of 40 cycles for a positive result. The correlation between the qRT-PCR results and the number of mapped reads (to NCBI RefSeq and study viral sequences) was evaluated by fitting a simple linear regression model for each sequencing depth. The R 2 and Pearson’s correlation coefficient were computed for each fitted line to compare the depth of sequencing. The limit of detection was estimated for each sample by estimating the Ct value corresponding to zero mapped reads. For influenza D results, qRT-PCR correlation was based on reads mapped to the conserved segment 6 (coding for polyprotein P42). Impact of sequencing depth on viral genome completeness The host sequences were filtered out of the subsampled set using the host read removal tool, Hostile, which included masking sequences of bovine bacteria and bovine viruses (Constantinides et al. 2023 ). Files generated in the previous mapping stage (.bam files) were used to determine the percentage coverage of each of the virus sequences using Samtools. The study workflow is presented in Fig. 1 . The percentage coverage of the genomes was plotted against the qRT-PCR results (Fig. 1 ). Numeric values were imported into R and visualized using the ggplot2 package (Wickham 2016 ). All computations were conducted on the high-performance computer at the University of Technology Sydney and the Elizabeth Macarthur Agricultural Institute cluster. Nucleotide differences between the NCBI RefSeq genome and study assembled genomes. To interpret the results, total nucleotide difference and differences across the genome between the study viruses and NCBI RefSeq were estimated using a similarity plot. The similarity plot was made using R with packages ape and Biostrings and plotted using gglot2 (Pagès et al. 2024 ; Team 2020 ; Wickham 2016 ; Paradis and Schliep 2019 ). The distance was calculated using K80 substitution model with a window size of 50 and step size of 25 for the sliding window. Results qRT-PCR results (Ct values) The Ct values combined for each of pools that were subsequently sequenced, are shown in Table 1 . Of the fifteen pools, all were positive to BNV with Ct values between 15.537-29.00. All but one pool was positive to IDV, similar to BCoV, and five pools were negative to BVDV. Table 1 Ct values for each of the five viruses in the metatranscriptomics pools. IDV = Influenza D virus, BNV = Bovine nidovirus, BVDV = Bovine Viral Diarrhea Virus, BCoV = Bovine Coronavirus. Ct Metagenome IDV BNV BVDV BCoV pool1 > 40 21.954 30.719 25.650 pool2 34.184 23.059 > 40 34.903 pool3 29.092 20.695 28.660 24.669 pool4 33.408 29.000 > 40 > 40 1 22.037 15.537 31.571 35.000 2 34.511 19.841 > 40 31.683 3 31.878 16.928 34.600 34.593 4 29.121 16.855 30.363 32.556 5 28.928 17.460 27.828 33.796 12 24.996 19.726 34.191 31.741 7 22.799 17.686 38.100 28.474 8 29.637 17.135 > 40 29.976 9 23.185 20.520 > 40 34.091 10 26.451 20.392 35.270 35.560 11 25.200 20.737 38.028 36.401 Impact of sequencing depth on mapped reads : Bovine nidovirus: The number of reads mapped to both NCBI RefSeq and study-specific viral sequences correlated with qRT-PCR Ct values, showing similar regression slopes across sequencing depths. BNV was detected in all samples (Fig. 2 A). At sequencing depth of 10 million reads, the predicted Ct value at whicn no reads would be detected (i.e., the x-intercept of the regression)was 43.56 cycles- suggesting greater sensitivity than qRT-PCR, which typically uses a threshold of 40 cycles. In contrast a 1 million reads, the corresponding Ct threshold was estimated at 38.58 cycles. The R 2 value for the reads mapped to the study-specific virus was consistent across depth (1M, 10M and 20M), with a value of 0.65 (Fig. 2 A). Bovine Coronavirus: There were no reads mapped to BCoV using sequencing depth 10k reads. One sample that was negative by qRT-PCR (pool4 Ct > 40) had no reads mapped to references at any depth with metagenomics sequencing. Th Ct values of the remaining 14 libraries ranged between 24.669–36.4. At sequencing depth = 1M there were no reads mapped to references in four qRT-PCR positive samples. At depth of 10M and 20M, all PCR positive samples had a minimum of 2 reads mapped to the references. There was an increasing slope for the fitted regression with higher sequencing depth, resulting in similar Ct values at the cutoff when no reads mapped (38.43, 40.54 and 41.23 for 1M, 10M and 20M respectively (Fig. 2 B). Influenza D virus (segment 6): At 20M sequencing read depth four samples had no reads mapped to NCBI RefSeq or study assembled viruses but were positive in the qRT-PCR (Ct = 29.64–34.5). Six qRT-PCR positive samples did not have reads mapped at 10M sequencing depth. At 1M, ten samples had no reads mapped to references. IDV was not detected in one of the pools (pool 1) by qRT-PCR, consistent with no reads mapped to any of the IDV genome segment. The regression slopes were higher for higher sequencing depths, with all showing a similar metagenome limit of detection (zero reads mapped to reference) of the qRT-PCR between Ct33.18 and 34.77. R 2 and Pearson correlation was higher for the regression fitted to the highest sequencing depth (Fig. 2 C). Overall, the sensitivity of metatranscriptomics was lower in IDV compared to other viruses. Bovine viral diarrhea virus-1: Similar to the other viruses, more reads were mapped to references at higher sequencing depth. Using the study viral genome, the slopes at the higher sequencing depth 10M and 20M were steeper compared to the slope at 1M sequencing depth. Five samples gave negative qRT-PCR results and these libraries did not have reads mapped to the study reference at 20M sequencing depth, however reads from these samples were mapped to the NCBI RefSeq. The estimated limit of metagenomic detection (0 mapped reads) was similar for all depths base on the regression: 39.58. 40.38, 40.88 for sequencing depths 1M, 10M, and 20M respectively when using the study assembled reference. However, at 20M, only one PCR positive sample had no reads mapped to the study assembled contigs, whereas two had no reads mapped at 10M and five samples at 1M. R 2 and Pearson correlation was similar for sequencing depths 10M and 20M. Using the NCBI RefSeq genome gave random mapped sequences across different Ct values (Fig. 2 D). To determine if any of the IDV segments are present in higher quantities, we estimated the number of reads mapped to study assemblies and NCBI reference sequences were estimated for each of the segments (Fig. 3 ). Overall, most segments had a similar number of mapped reads. Completeness of genomes at different sequencing depth (coverage breadth) The completeness of the viral genome obtained with reference mapping at different sequencing depths was not always consistent with the viral RNA measured by qRT-PCR. Bovine Coronavirus had a high genome sequence coverage with Ct values 30, except for one Ct = 35 sample where almost full coverage was recovered (Fig. 4 ). For Bovine nidovirus, all samples had Ct < 30 and the coverage of near complete genome was achieved for all samples with sequencing depths 10M and 20M. With the lowest Ct value = 15.54 near complete genome was obtained with sequencing depth as low as 10k using the study assembly as the reference (Fig. 4 ). For IDV high coverage was obtained with Ct < 26 with coverage depths of 10M and 20M and drops to less < 25% coverage after Ct = 30 with any of the sequencing depths (Fig. 4 ). For Bovine Viral Diarrhea virus-1 two different study assembled reference genomes were used to assess coverage breadth (two field isolates of BVDV-1a and BVDV-1c were present in the sample). Of the two samples with Ct values < 30 only one had high coverage breadth (94.44%). Of the eight remaining positive samples, only one (Ct = 30.36) had a high coverage breadth (79.0%), while the coverage depth of the remaining ones ranged from 1 and 8.3% (Fig. 4 ). Impact of the reference sequence in mapped reads and genome coverage : Bovine coronavirus: Due to the high similarity of the study sequence and the study assembly (Fig. 5 ), there were no differences between the identification using mapped reads across different sequencing lengths and qRT-PCR Ct (Fig. 2 ). The genome coverage of reads using both sequences as reference for mapping did not change when using the study sequence or the NCBI reference (Fig. 4 ). Bovine nidovirus: The ability to map reads to a reference genome was similar when using either the NCBI RefSeq or the study-assembled viral genome with a strong agreement across samples where at least one read was mapped to the reference. Comparable correlations between mapped reads and qRT-PCR Ct values were observed regardless of the reference genome used (Fig. 2 B). With respect to coverage breadth, a maximum of approximately 75% was achieved when using NCBI RefSeq (Fig. 4 ). The portion of the genome where reads weren’t mapped correspond to a region of higher divergence between the RefSeq and the study-assembled genome (Fig. 6). Influenza D virus: Similar to BNV and BCoV, segment 6 of IDV assembled in this study and the NCBI RefSeq had a high similarity (Fig. 5 ). Similar correlation with mapped reads and genome breadth of segment 6 was obtained regardless of the reference genome used. Bovine viral diarrhea virus-1: While there were regression slopes for reads mapped to the study assembled virus, the reads mapped to the NCBI BVDV-1 reference slopes were almost flat, with no prediction or correlation to the PCR results. Five samples were negative on PCR, however, all samples had similar number of mapped reads to the references. The genetic distance of the study sequence and the NCBI reference was ~ 0.2 across most of the genome (Fig. 5 ), which may account for the low percentage coverage of read mapping (Fig. 4 ). Discussion This study evaluated the accuracy of metatranscriptomic viral detection and genome recovery of RNA viruses in comparison to standard molecular diagnostic assays (qRT-PCR). Specifically, we assessed the impact of sequencing depth and reference genome selection on the detection of four bovine respiratory RNA viruses. Our findings highlight that sequencing depth plays a critical role in the sensitivity of viral detection. For non-segmented viruses such as BCoV, BNV, and BVDV-1, sequencing depths of 10 million reads or higher achieved detection thresholds comparable to qRT-PCR. However, sensitivity was reduced for IDV, with no reads mapped in samples with Ct values greater than 37.7, even at the highest sequencing depth. These results align with previous studies; for example, Zhang et al. 2020 (Zhang et al. 2020 ) reported metagenomic detection of IDV in 14 out of 17 samples with Ct values 31. In their study, sensitivity and specificity were estimated at 28.3% and 98.9%, respectively, using approximately 300,000–400,000 reads per sample (Zhang et al. 2020 ). In our study, we found that even with a high sequencing depth of 20M reads four samples had no reads mapped to NCBI RefSeq or study virus contigs but were positive by qRT-PCR (Ct values 29.64–34.5). Reference genome selection had an impact on detection and genome recovery, particularly for BVDV-1. Mapping to the study-assembled genome detected more reads than the NCBI RefSeq, likely due to sequence divergence with the field strain. This highlights the need for curated, population-specific viral genome databases to improve metagenomic diagnostic performance. In contrast, detection of BCoV, BNV, and IDV was consistent regardless of the reference due to higher nucleotide similarity (> 80%) between the NCBI RefSeq and the study assembled viral genome. In the case of BVDV-1, we observed that five samples that were negative by qRT-PCR still had mapped reads to both the NCBI RefSeq and the study-assembled references. One possible explanation for this observation is low-level index swapping (also known as index hopping) during sequencing, which can occasionally lead to false-positive signals, particularly in high-throughput platforms using patterned flow cells (Costello et al. 2018 ). While we cannot fully rule out this possibility, the overall abundance of BVDV-1 reads in these samples was relatively low, and we did not observe similar cross-sample contamination for the other viruses assessed. In those cases, PCR-negative samples consistently showed no mapped reads. This suggests that if index swapping did occur, its impact was limited and virus-specific, possibly influenced by the amount of viral RNA present in the original samples. Genome completeness (coverage breadth) varied by virus and sequencing depth. Near-complete genomes were recovered for BCoV in samples with lower Ct values regardless of the reference, while coverage was more limited for BNV, especially when relying on reference genomes with partial divergence. These results are relevant not only for detection but also for downstream applications such as molecular epidemiology and strain-level characterization. A practical challenge in implementing metagenomics sequencing is determining the appropriate sequencing depth. Unlike whole genome sequencing of isolates—where depth can be estimated based on genome size—the complexity of metagenomic samples and variability in host background make it difficult to define optimal sequencing parameters a priori (Valiente-Mullor et al. 2021 ). Our findings show that samples with Ct values below 30 can be good candidates for recovering > 75% of a viral genome given appropriate sequencing depth. A clear understanding of the sensitivity of metatranscriptomics is essential for its diagnostic interpretation. Comparison with an established method such as qRT-PCR helps contextualize the performance of this non-enriched approach. However, implementing metagenomics as a routine diagnostic tool requires further optimization of sample processing and sequencing protocols. Just as qRT-PCR workflows have been standardized over time (e.g., fixed cycle thresholds), sequencing-based workflows must also be calibrated to identify cost-effective and diagnostically meaningful depth. Further development of robust bioinformatics pipelines is also necessary to automate and improve the accuracy of viral detection. Our findings reinforce the need for tailored, population-specific viral reference databases for veterinary applications. Although total RNA sequencing provides an unbiased method for virus detection, it requires higher sequencing depth and robust computational infrastructure. The decreasing costs of high-throughput sequencing have made metagenomics more accessible; however, data processing and storage remain significant barriers for implementation in large-scale or routine veterinary and agricultural diagnostics. Additionally, sample collection, preservation, and RNA extraction quality can strongly influence sensitivity. Targeted capture and enrichment methods have shown promise in improving viral detection sensitivity, but they are currently optimized for human diagnostics and are not widely adapted for veterinary pathogens (Buddle et al. 2024 ). Together, these findings support the potential of metatranscriptomics to enhance pathogen detection in veterinary diagnostics, particularly when conventional methods fall short of identifying novel pathogens. However, realizing this potential will require continued investment in reference curated database development, optimization of sequencing approaches, and bioinformatic tools for routine diagnostic analysis. By addressing these gaps, metatranscriptomic approaches can move closer to practical application in the field, offering a powerful complement to established molecular diagnostics. Declarations Acknowledgements The authors would like to thank the eResearch Team from the Computational Research Support Unit (CRSU) at the University of Technology Sydney (UTS) for his invaluable support in providing the computational infrastructure and guidance for the bioinformatics analyses conducted in this study. Statement of Animal Ethics The cattle sampling procedure used in this study was approved by the University of Technology Sydney Animal Care and Ethics Committee (Approval No. ETH19-3407). Conflict of Interest Statement The authors declare no conflict of interest. Funding Financial support for this project was provided by UTS and the Australian Centre for Genomic Epidemiological Microbiology (AusGEM), a collaborative partnership between the NSW Department of Primary Industries and the University of Technology Sydney. References Govender, K.N., Street, T.L., Sanderson, N.D. & Eyre, D.W. Metagenomic Sequencing as a Pathogen-Agnostic Clinical Diagnostic Tool for Infectious Diseases: a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies. J Clin Microbiol 59 ,e0291620 (2021). Liu, J., Zhang, Q., Dong, Y.Q., Yin, J. & Qiu, Y.Q. Diagnostic accuracy of metagenomic next-generation sequencing in diagnosing infectious diseases: a meta-analysis. Sci Rep 12 ,21032 (2022). Neyton, L.P.A., Langelier, C.R. & Calfee, C.S. Metagenomic Sequencing in the ICU for Precision Diagnosis of Critical Infectious Illnesses. Crit Care 27 ,90 (2023). Shakya, M., Lo, C.C. & Chain, P.S.G. Advances and Challenges in Metatranscriptomic Analysis. Front Genet 10 ,904 (2019). Cobbin, J.C., Charon, J., Harvey, E., Holmes, E.C. & Mahar, J.E. Current challenges to virus discovery by meta-transcriptomics. Curr Opin Virol 51 ,48-55 (2021). Brito, B.P. et al. Expanding the range of the respiratory infectome in Australian feedlot cattle with and without respiratory disease using metatranscriptomics. Microbiome 11 ,158 (2023). Bushnell, B. BBMap: A Fast, Accurate, Splice-Aware Aligner. In: Institute, J.G., editor.; 2014. Md, V., Misra, S., Li, H. & Aluru, S. Efficient Architecture-Aware Acceleration of BWA-MEM for Multicore Systems. IEEE Parallel and Distributed Processing Symposium (IPDPS) ; 2019. Danecek, P. et al. Twelve years of SAMtools and BCFtools. Gigascience 10 (2021). Constantinides, B., Hunt, M. & Crook, D.W. Hostile: accurate decontamination of microbial host sequences. Bioinformatics 39 (2023). Wickham, H. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag; 2016. Pagès, H., Aboyoun, P., Gentleman, R. & DebRoy, S. Biostrings: Efficient manipulation of biological strings. R package version 2.74.1. 2024. Team, R.D.C. R: A language and environment for statistical computing. In: Computing, R.F.f.S., editor.; 2020. Paradis, E. & Schliep, K. ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35 ,526-528 (2019). Zhang, M. et al. Assessment of Metagenomic Sequencing and qPCR for Detection of Influenza D Virus in Bovine Respiratory Tract Samples. Viruses 12 (2020). Costello, M. et al. Characterization and remediation of sample index swaps by non-redundant dual indexing on massively parallel sequencing platforms. BMC Genomics 19 ,332 (2018). Valiente-Mullor, C. et al. One is not enough: On the effects of reference genome for the mapping and subsequent analyses of short-reads. PLoS Comput Biol 17 ,e1008678 (2021). Buddle, S. et al. Evaluating metagenomics and targeted approaches for diagnosis and surveillance of viruses. Genome Med 16 ,111 (2024). Additional Declarations No competing interests reported. Supplementary Files TableS1.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7071637","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":484385609,"identity":"97185548-6e0e-410c-a418-e15092268af7","order_by":0,"name":"Barbara Brito","email":"data:image/png;base64,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","orcid":"","institution":"University of Technology Sydney","correspondingAuthor":true,"prefix":"","firstName":"Barbara","middleName":"","lastName":"Brito","suffix":""},{"id":484385610,"identity":"76891589-6192-4d18-99ae-01427eed415e","order_by":1,"name":"Melinda Frost","email":"","orcid":"","institution":"Elizabeth Macarthur Agricultural Institute","correspondingAuthor":false,"prefix":"","firstName":"Melinda","middleName":"","lastName":"Frost","suffix":""},{"id":484385611,"identity":"b57717ea-46be-4c22-8a41-6e47f2260de7","order_by":2,"name":"John Webster","email":"","orcid":"","institution":"Elizabeth Macarthur Agricultural Institute","correspondingAuthor":false,"prefix":"","firstName":"John","middleName":"","lastName":"Webster","suffix":""},{"id":484385612,"identity":"2d822586-b1e9-4baf-a7e5-5781377ec04a","order_by":3,"name":"Joyce To","email":"","orcid":"","institution":"University of Technology Sydney","correspondingAuthor":false,"prefix":"","firstName":"Joyce","middleName":"","lastName":"To","suffix":""},{"id":484385613,"identity":"7e79bb32-f68c-44f0-bfca-5db53b511ac2","order_by":4,"name":"Peter Kirkland","email":"","orcid":"","institution":"Elizabeth Macarthur Agricultural Institute","correspondingAuthor":false,"prefix":"","firstName":"Peter","middleName":"","lastName":"Kirkland","suffix":""}],"badges":[],"createdAt":"2025-07-08 07:08:29","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7071637/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7071637/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":87229524,"identity":"4007469f-4b97-47f3-865e-0cadad42c9ae","added_by":"auto","created_at":"2025-07-21 18:14:01","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":310361,"visible":true,"origin":"","legend":"\u003cp\u003eOverview of the study workflow for viral detection and genome analysis. The workflow begins with total RNA sequencing of 15 libraries derived from nasal swabs. The samples were screened for five viruses—Bovine Coronavirus (BCoV), Bovine nidovirus (BNV), Bovine Viral Diarrhea Virus 1 (BVDV-1) and Influenza D Virus (IDV)—using qRT-PCR. The reads from the sequenced transcriptomes were trimmed discarding the sequences with low quality and then subsampled to depths of 20M, 10M, and 1M reads for downstream analysis. The reads were mapped to reference genomes and study-assembled viral genomes using BWA-MEM2 to assess alignment metrics and viral genome completeness.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7071637/v1/436e121b56e9d93bb535da36.png"},{"id":87229525,"identity":"af4b05d0-38dd-403a-b83a-c4ef24ce86e2","added_by":"auto","created_at":"2025-07-21 18:14:01","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":916360,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelations between the mapped reads and RT-PCR cycle threshold (Ct) log10 values across sequencing depths (1M, 10M, 20M) and reference genomes. The scatter plots depict the association between reads mapped to the study virus and the NCBI reference sequence and Ct values for Bovine nidovirus (2A), Bovine Coronavirus (2B), Influenza D Virus (C) and Bovine Viral Diarrhea Virus-1 (2D). The statistics of the fitted regression and Pearson correlation coefficient are displayed in the bottom tables with the estimated Ct (log) values at 0 mapped reads.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7071637/v1/a7db0bf4996459c64ca4a199.png"},{"id":87228872,"identity":"04c0963a-4bf0-4b9b-ad2b-16548564ea59","added_by":"auto","created_at":"2025-07-21 18:06:01","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":229887,"visible":true,"origin":"","legend":"\u003cp\u003eNumber of reads mapped to IDV segments from the study assembled genomes and NCBI reference at 10k, 1M, 10M and 20M sequencing depth transcriptomes.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7071637/v1/9bcce0a5cd3eae30d1aee9a1.png"},{"id":87228870,"identity":"d45fa7be-6c69-41b7-906a-45e0fd7ecc0d","added_by":"auto","created_at":"2025-07-21 18:06:01","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":358200,"visible":true,"origin":"","legend":"\u003cp\u003eGenome completeness of Bovine Viral Diarrhea-1, Bovine coronavirus, Bovine nidovirus and IDV different sequencing length.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7071637/v1/a6087a811f43881ffb82ad11.png"},{"id":87228873,"identity":"cd031e5f-c42f-4a99-aee2-dfe00b9ebf6d","added_by":"auto","created_at":"2025-07-21 18:06:01","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":527679,"visible":true,"origin":"","legend":"\u003cp\u003eSimilarity plot of NCBI reference sequences and study sequences. There is a high similarity between the study sequences and NCBI references in the BCoV genome and IDV segment 6. Bovine viral diarrhea virus-1 reference similarities are low at ~20-40% across most of the genome.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7071637/v1/c0ecca15b1b5d3954f2a13f3.png"},{"id":88173760,"identity":"15d7d1d4-7e99-43ae-8afa-6e87a0208c47","added_by":"auto","created_at":"2025-08-02 22:46:24","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3335093,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7071637/v1/8e2b9071-aa40-49d5-b571-e86b56da52a8.pdf"},{"id":87228869,"identity":"487a4168-aff0-439f-bb22-c2e7aebd81ec","added_by":"auto","created_at":"2025-07-21 18:06:01","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":16879,"visible":true,"origin":"","legend":"","description":"","filename":"TableS1.docx","url":"https://assets-eu.researchsquare.com/files/rs-7071637/v1/b15d6cd753cc4a49d01b408a.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Detecting RNA Viruses in Cattle: Effects of Sequencing Depth and Sequence References in Metatranscriptomics","fulltext":[{"header":"Introduction","content":"\u003cp\u003eMetatranscriptomics, the untargeted sequencing of RNA from clinical samples, is increasingly used for the detection and characterization of RNA viruses. This approach enables the identification of both known and novel viruses without prior knowledge of their genomic sequences, making it particularly useful for outbreak investigations and surveillance.\u003c/p\u003e\u003cp\u003eWhile metatranscriptomics offers a comprehensive approach to viral detection, its diagnostic accuracy compared to conventional methods remains insufficiently characterized. Meta-analyses of untargeted metagenomic diagnostics in human medicine—often evaluating blood, cerebrospinal fluid, orthopedic, and respiratory samples—have reported overall sensitivities ranging from 0.75 to 0.90 and specificities between 0.67 and 0.96, depending on sample type (Govender et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Liu et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Neyton et al. \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). However, these evaluations have been limited to human clinical contexts, and comparable studies in veterinary medicine—particularly for RNA viruses in livestock—are largely lacking.\u003c/p\u003e\u003cp\u003eAlthough metagenomic approaches are increasingly being explored for diagnostic purposes, their integration into routine veterinary investigations remains limited. As a result, there is a pressing need for systematic evaluations that benchmark their performance against established diagnostic standards such as quantitative reverse transcription PCR (qRT-PCR). While qRT-PCR remains the gold standard due to its high sensitivity and specificity, it is inherently limited to the detection of known viral targets. In contrast, metatranscriptomics enables untargeted detection of both expected and novel RNA viruses, but its diagnostic accuracy—particularly in relation to viral genome divergence and sequencing yield per sample—has not been rigorously assessed.\u003c/p\u003e\u003cp\u003eA key technical consideration is how the quantity of sequencing data (sequencing depth) influences the sensitivity of detection. Although high-throughput sequencing platforms can produce substantial read counts, practical and economic constraints often limit the number of reads per sample. Consequently, there is a need to determine how sequencing depth affects detection sensitivity, particularly across a range of viral loads.\u003c/p\u003e\u003cp\u003eAnother important consideration in metatranscriptomic analysis is the choice of reference genome. While some bioinformatic pipelines incorporate de novo assembly and taxonomic classification of contigs, these approaches are computationally intensive and more difficult to automate compared to direct mapping to reference genomes (Shakya et al. \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). However, the accuracy of read-mapping–based detection depends heavily on the sequence similarity between the virus present in the sample and the chosen reference genome (Cobbin et al. \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). For many veterinary pathogens—particularly RNA viruses—there is limited representation in public databases, and field strains often differ significantly from available references. The extent to which this genomic divergence impacts detection sensitivity remains poorly characterized.\u003c/p\u003e\u003cp\u003eTo address these gaps, this study evaluates the diagnostic performance of metatranscriptomics for detecting four clinically relevant bovine respiratory RNA viruses—Bovine Coronavirus (BCoV), Bovine nidovirus (BNV), Influenza D Virus (IDV), and Bovine Viral Diarrhea Virus-1 (BVDV-1). Using pooled nasal and nasopharyngeal samples, we compared mapped read counts against qRT-PCR cycle threshold (Ct) values across subsampled sequencing depths (20M, 10M, and 1M reads). We further assessed how reference genome choice—comparing NCBI RefSeq entries with viral genomes assembled from the same dataset—affects detection and genome coverage. This study aims to provide critical insights for optimizing metagenomic approaches in veterinary diagnostics.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cem\u003eSamples and data\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe study samples included nasopharyngeal and nasal swabs collected in transport medium designed to preserve nucleic acid (Copan\u0026reg; ENAT and Copan\u0026reg; UTM) from cattle in a feedlot in New South Wales, Australia. The libraries used in this study are detailed in Brito et al., \u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e (Brito et al. \u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e). In brief, pooled RNA extracted from nasal swabs from animals with or without bovine respiratory disease were analysed. RNA extraction was performed using the RNeasy Plus Micro Kit and RNeasy Plus Mini Kit (Qiagen). Libraries were prepared using the TruSeq Stranded Total RNA Human/Mouse/Rat Kit (Illumina) and sequencing was conducted on a NovaSeq S1 (Illumina) platform at the Australian Genome Research Facility. The total read count per sample and corresponding Bioproject details are provided in Table S1.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eRT-qPCR\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFour quantitative real-time PCR (qRT-PCR) targeting Bovine nidovirus (BNV), Bovine Coronavirus (BCoV), Influenza D virus (IDV), and Bovine Viral Diarrhea-1 (BVDV) were conducted on non-pooled samples. These samples were pooled before next generation sequencing library preparation as described above. The Ct values for each of the 15 pools was estimated from the combined values of the individual samples that comprised each pool. To estimate the cycle threshold (Ct) values equivalent to the pooled libraries, the total copy number of each contributing sample was calculated using a 95% efficiency slope equation, Ct\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;3.45x\u0026thinsp;+\u0026thinsp;40. The cumulative copy number from all samples was subsequently converted back into a corresponding Ct value for each pooled library.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eImpact of sequencing depth and reference sequences in viral detection\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe raw sequences were initially quality trimmed using BBDuk from BBTools 38.87 (Bushnell \u003cspan class=\"CitationRef\"\u003e2014\u003c/span\u003e). To evaluate how the depth and reference sequence selection influence viral detection, each of the QC trimmed metatranscriptomes were randomly subsampled to approximate different sequencing depths. Subsamples of 20M, 10M, and 1M reads were generated from each library using seqtk sample (with the -s100 seed option) (github.com/lh3/seqtk).\u003c/p\u003e\n\u003cp\u003eThe subsampled reads were mapped to both, the viruses assembled in this study and the viral species RefSeq from NCBI for Bovine nidovirus (NC_027199.1), Bovine Coronavirus (NC_003045.1), Influenza D virus (NC_036615-21) and Bovine Viral Diarrhea-1 (NC_001461.1) using the BWA-MEM2 aligner (Md et al. \u003cspan class=\"CitationRef\"\u003e2019\u003c/span\u003e). A mapping quality threshold of an alignment score\u0026thinsp;\u0026gt;\u0026thinsp;=\u0026thinsp;100 was applied to retain high-quality mapped contigs and Samtools was used to compute the mapping statistics and read coverage for each reference (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e) (Danecek et al. \u003cspan class=\"CitationRef\"\u003e2021\u003c/span\u003e). For the segmented virus Influenza D, mapped reads were obtained for all seven segments to determine if there were major differences in read mapping by segment.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCorrelation between mapped reads and qRT-PCR, and differences across reference genomes\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eqRT-PCR was performed for the four viruses mentioned above, using a limit of 40 cycles for a positive result. The correlation between the qRT-PCR results and the number of mapped reads (to NCBI RefSeq and study viral sequences) was evaluated by fitting a simple linear regression model for each sequencing depth. The R\u003csup\u003e2\u003c/sup\u003e and Pearson\u0026rsquo;s correlation coefficient were computed for each fitted line to compare the depth of sequencing. The limit of detection was estimated for each sample by estimating the Ct value corresponding to zero mapped reads. For influenza D results, qRT-PCR correlation was based on reads mapped to the conserved segment 6 (coding for polyprotein P42).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eImpact of sequencing depth on viral genome completeness\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe host sequences were filtered out of the subsampled set using the host read removal tool, Hostile, which included masking sequences of bovine bacteria and bovine viruses (Constantinides et al. \u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e). Files generated in the previous mapping stage (.bam files) were used to determine the percentage coverage of each of the virus sequences using Samtools. The study workflow is presented in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e. The percentage coverage of the genomes was plotted against the qRT-PCR results (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\n\u003cp\u003eNumeric values were imported into R and visualized using the ggplot2 package (Wickham \u003cspan class=\"CitationRef\"\u003e2016\u003c/span\u003e). All computations were conducted on the high-performance computer at the University of Technology Sydney and the Elizabeth Macarthur Agricultural Institute cluster.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNucleotide differences between the NCBI RefSeq genome and study assembled genomes.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eTo interpret the results, total nucleotide difference and differences across the genome between the study viruses and NCBI RefSeq were estimated using a similarity plot. The similarity plot was made using R with packages ape and Biostrings and plotted using gglot2 (Pag\u0026egrave;s et al. \u003cspan class=\"CitationRef\"\u003e2024\u003c/span\u003e; Team \u003cspan class=\"CitationRef\"\u003e2020\u003c/span\u003e; Wickham \u003cspan class=\"CitationRef\"\u003e2016\u003c/span\u003e; Paradis and Schliep \u003cspan class=\"CitationRef\"\u003e2019\u003c/span\u003e). The distance was calculated using K80 substitution model with a window size of 50 and step size of 25 for the sliding window.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cb\u003eqRT-PCR results (Ct values)\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe Ct values combined for each of pools that were subsequently sequenced, are shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Of the fifteen pools, all were positive to BNV with Ct values between 15.537-29.00. All but one pool was positive to IDV, similar to BCoV, and five pools were negative to BVDV.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eCt values for each of the five viruses in the metatranscriptomics pools. IDV\u0026thinsp;=\u0026thinsp;Influenza D virus, BNV\u0026thinsp;=\u0026thinsp;Bovine nidovirus, BVDV\u0026thinsp;=\u0026thinsp;Bovine Viral Diarrhea Virus, BCoV\u0026thinsp;=\u0026thinsp;Bovine Coronavirus.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colspan=\"5\" nameend=\"c6\" namest=\"c2\"\u003e\u003cp\u003eCt\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMetagenome\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eIDV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eBNV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eBVDV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eBCoV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003epool1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;40\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e21.954\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e30.719\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e25.650\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003epool2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e34.184\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e23.059\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;40\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e34.903\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003epool3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e29.092\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20.695\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28.660\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e24.669\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003epool4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e33.408\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e29.000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;40\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;40\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e22.037\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15.537\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e31.571\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e35.000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e34.511\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e19.841\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;40\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e31.683\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e31.878\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e16.928\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e34.600\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e34.593\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e29.121\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e16.855\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e30.363\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e32.556\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e28.928\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17.460\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e27.828\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e33.796\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e12\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e24.996\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e19.726\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e34.191\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e31.741\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e22.799\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17.686\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e38.100\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e28.474\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e29.637\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17.135\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;40\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e29.976\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e9\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e23.185\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20.520\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;40\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e34.091\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e26.451\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20.392\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e35.270\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e35.560\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e11\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e25.200\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20.737\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e38.028\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e36.401\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cem\u003eImpact of sequencing depth on mapped reads\u003c/em\u003e:\u003c/p\u003e\u003cp\u003eBovine nidovirus: The number of reads mapped to both NCBI RefSeq and study-specific viral sequences correlated with qRT-PCR Ct values, showing similar regression slopes across sequencing depths. BNV was detected in all samples (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). At sequencing depth of 10\u0026nbsp;million reads, the predicted Ct value at whicn no reads would be detected (i.e., the x-intercept of the regression)was 43.56 cycles- suggesting greater sensitivity than qRT-PCR, which typically uses a threshold of 40 cycles. In contrast a 1\u0026nbsp;million reads, the corresponding Ct threshold was estimated at 38.58 cycles. The R\u003csup\u003e2\u003c/sup\u003e value for the reads mapped to the study-specific virus was consistent across depth (1M, 10M and 20M), with a value of 0.65 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA).\u003c/p\u003e\u003cp\u003eBovine Coronavirus: There were no reads mapped to BCoV using sequencing depth 10k reads. One sample that was negative by qRT-PCR (pool4 Ct\u0026thinsp;\u0026gt;\u0026thinsp;40) had no reads mapped to references at any depth with metagenomics sequencing. Th Ct values of the remaining 14 libraries ranged between 24.669\u0026ndash;36.4. At sequencing depth\u0026thinsp;=\u0026thinsp;1M there were no reads mapped to references in four qRT-PCR positive samples. At depth of 10M and 20M, all PCR positive samples had a minimum of 2 reads mapped to the references. There was an increasing slope for the fitted regression with higher sequencing depth, resulting in similar Ct values at the cutoff when no reads mapped (38.43, 40.54 and 41.23 for 1M, 10M and 20M respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB).\u003c/p\u003e\u003cp\u003eInfluenza D virus (segment 6): At 20M sequencing read depth four samples had no reads mapped to NCBI RefSeq or study assembled viruses but were positive in the qRT-PCR (Ct\u0026thinsp;=\u0026thinsp;29.64\u0026ndash;34.5). Six qRT-PCR positive samples did not have reads mapped at 10M sequencing depth. At 1M, ten samples had no reads mapped to references. IDV was not detected in one of the pools (pool 1) by qRT-PCR, consistent with no reads mapped to any of the IDV genome segment. The regression slopes were higher for higher sequencing depths, with all showing a similar metagenome limit of detection (zero reads mapped to reference) of the qRT-PCR between Ct33.18 and 34.77. R\u003csup\u003e2\u003c/sup\u003e and Pearson correlation was higher for the regression fitted to the highest sequencing depth (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). Overall, the sensitivity of metatranscriptomics was lower in IDV compared to other viruses.\u003c/p\u003e\u003cp\u003eBovine viral diarrhea virus-1: Similar to the other viruses, more reads were mapped to references at higher sequencing depth. Using the study viral genome, the slopes at the higher sequencing depth 10M and 20M were steeper compared to the slope at 1M sequencing depth. Five samples gave negative qRT-PCR results and these libraries did not have reads mapped to the study reference at 20M sequencing depth, however reads from these samples were mapped to the NCBI RefSeq.\u0026nbsp;The estimated limit of metagenomic detection (0 mapped reads) was similar for all depths base on the regression: 39.58. 40.38, 40.88 for sequencing depths 1M, 10M, and 20M respectively when using the study assembled reference. However, at 20M, only one PCR positive sample had no reads mapped to the study assembled contigs, whereas two had no reads mapped at 10M and five samples at 1M. R\u003csup\u003e2\u003c/sup\u003e and Pearson correlation was similar for sequencing depths 10M and 20M. Using the NCBI RefSeq genome gave random mapped sequences across different Ct values (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo determine if any of the IDV segments are present in higher quantities, we estimated the number of reads mapped to study assemblies and NCBI reference sequences were estimated for each of the segments (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Overall, most segments had a similar number of mapped reads.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cem\u003eCompleteness of genomes at different sequencing depth (coverage breadth)\u003c/em\u003e\u003c/p\u003e\u003cp\u003eThe completeness of the viral genome obtained with reference mapping at different sequencing depths was not always consistent with the viral RNA measured by qRT-PCR.\u003c/p\u003e\u003cp\u003eBovine Coronavirus had a high genome sequence coverage with Ct values\u0026thinsp;\u0026lt;\u0026thinsp;30 at 10M and 20M sequencing depth, and dropped at Ct\u0026thinsp;\u0026gt;\u0026thinsp;30, except for one Ct\u0026thinsp;=\u0026thinsp;35 sample where almost full coverage was recovered (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eFor Bovine nidovirus, all samples had Ct\u0026thinsp;\u0026lt;\u0026thinsp;30 and the coverage of near complete genome was achieved for all samples with sequencing depths 10M and 20M. With the lowest Ct value\u0026thinsp;=\u0026thinsp;15.54 near complete genome was obtained with sequencing depth as low as 10k using the study assembly as the reference (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eFor IDV high coverage was obtained with Ct\u0026thinsp;\u0026lt;\u0026thinsp;26 with coverage depths of 10M and 20M and drops to less \u0026lt;\u0026thinsp;25% coverage after Ct\u0026thinsp;=\u0026thinsp;30 with any of the sequencing depths (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eFor Bovine Viral Diarrhea virus-1 two different study assembled reference genomes were used to assess coverage breadth (two field isolates of BVDV-1a and BVDV-1c were present in the sample). Of the two samples with Ct values\u0026thinsp;\u0026lt;\u0026thinsp;30 only one had high coverage breadth (94.44%). Of the eight remaining positive samples, only one (Ct\u0026thinsp;=\u0026thinsp;30.36) had a high coverage breadth (79.0%), while the coverage depth of the remaining ones ranged from 1 and 8.3% (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cem\u003eImpact of the reference sequence in mapped reads and genome coverage\u003c/em\u003e:\u003c/p\u003e\u003cp\u003eBovine coronavirus: Due to the high similarity of the study sequence and the study assembly (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e), there were no differences between the identification using mapped reads across different sequencing lengths and qRT-PCR Ct (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The genome coverage of reads using both sequences as reference for mapping did not change when using the study sequence or the NCBI reference (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eBovine nidovirus: The ability to map reads to a reference genome was similar when using either the NCBI RefSeq or the study-assembled viral genome with a strong agreement across samples where at least one read was mapped to the reference. Comparable correlations between mapped reads and qRT-PCR Ct values were observed regardless of the reference genome used (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). With respect to coverage breadth, a maximum of approximately 75% was achieved when using NCBI RefSeq (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). The portion of the genome where reads weren\u0026rsquo;t mapped correspond to a region of higher divergence between the RefSeq and the study-assembled genome (Fig.\u0026nbsp;6).\u003c/p\u003e\u003cp\u003eInfluenza D virus: Similar to BNV and BCoV, segment 6 of IDV assembled in this study and the NCBI RefSeq had a high similarity (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Similar correlation with mapped reads and genome breadth of segment 6 was obtained regardless of the reference genome used.\u003c/p\u003e\u003cp\u003eBovine viral diarrhea virus-1: While there were regression slopes for reads mapped to the study assembled virus, the reads mapped to the NCBI BVDV-1 reference slopes were almost flat, with no prediction or correlation to the PCR results. Five samples were negative on PCR, however, all samples had similar number of mapped reads to the references. The genetic distance of the study sequence and the NCBI reference was ~\u0026thinsp;0.2 across most of the genome (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e), which may account for the low percentage coverage of read mapping (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study evaluated the accuracy of metatranscriptomic viral detection and genome recovery of RNA viruses in comparison to standard molecular diagnostic assays (qRT-PCR). Specifically, we assessed the impact of sequencing depth and reference genome selection on the detection of four bovine respiratory RNA viruses.\u003c/p\u003e\u003cp\u003eOur findings highlight that sequencing depth plays a critical role in the sensitivity of viral detection. For non-segmented viruses such as BCoV, BNV, and BVDV-1, sequencing depths of 10\u0026nbsp;million reads or higher achieved detection thresholds comparable to qRT-PCR. However, sensitivity was reduced for IDV, with no reads mapped in samples with Ct values greater than 37.7, even at the highest sequencing depth. These results align with previous studies; for example, Zhang et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2020\u003c/span\u003e (Zhang et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) reported metagenomic detection of IDV in 14 out of 17 samples with Ct values\u0026thinsp;\u0026lt;\u0026thinsp;31 using short-read sequencing, while detection dropped significantly for samples with Ct\u0026thinsp;\u0026gt;\u0026thinsp;31. In their study, sensitivity and specificity were estimated at 28.3% and 98.9%, respectively, using approximately 300,000\u0026ndash;400,000 reads per sample (Zhang et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). In our study, we found that even with a high sequencing depth of 20M reads four samples had no reads mapped to NCBI RefSeq or study virus contigs but were positive by qRT-PCR (Ct values 29.64\u0026ndash;34.5).\u003c/p\u003e\u003cp\u003eReference genome selection had an impact on detection and genome recovery, particularly for BVDV-1. Mapping to the study-assembled genome detected more reads than the NCBI RefSeq, likely due to sequence divergence with the field strain. This highlights the need for curated, population-specific viral genome databases to improve metagenomic diagnostic performance. In contrast, detection of BCoV, BNV, and IDV was consistent regardless of the reference due to higher nucleotide similarity (\u0026gt;\u0026thinsp;80%) between the NCBI RefSeq and the study assembled viral genome.\u003c/p\u003e\u003cp\u003eIn the case of BVDV-1, we observed that five samples that were negative by qRT-PCR still had mapped reads to both the NCBI RefSeq and the study-assembled references. One possible explanation for this observation is low-level index swapping (also known as index hopping) during sequencing, which can occasionally lead to false-positive signals, particularly in high-throughput platforms using patterned flow cells (Costello et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). While we cannot fully rule out this possibility, the overall abundance of BVDV-1 reads in these samples was relatively low, and we did not observe similar cross-sample contamination for the other viruses assessed. In those cases, PCR-negative samples consistently showed no mapped reads. This suggests that if index swapping did occur, its impact was limited and virus-specific, possibly influenced by the amount of viral RNA present in the original samples.\u003c/p\u003e\u003cp\u003eGenome completeness (coverage breadth) varied by virus and sequencing depth. Near-complete genomes were recovered for BCoV in samples with lower Ct values regardless of the reference, while coverage was more limited for BNV, especially when relying on reference genomes with partial divergence. These results are relevant not only for detection but also for downstream applications such as molecular epidemiology and strain-level characterization.\u003c/p\u003e\u003cp\u003eA practical challenge in implementing metagenomics sequencing is determining the appropriate sequencing depth. Unlike whole genome sequencing of isolates\u0026mdash;where depth can be estimated based on genome size\u0026mdash;the complexity of metagenomic samples and variability in host background make it difficult to define optimal sequencing parameters \u003cem\u003ea priori\u003c/em\u003e (Valiente-Mullor et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Our findings show that samples with Ct values below 30 can be good candidates for recovering\u0026thinsp;\u0026gt;\u0026thinsp;75% of a viral genome given appropriate sequencing depth.\u003c/p\u003e\u003cp\u003eA clear understanding of the sensitivity of metatranscriptomics is essential for its diagnostic interpretation. Comparison with an established method such as qRT-PCR helps contextualize the performance of this non-enriched approach. However, implementing metagenomics as a routine diagnostic tool requires further optimization of sample processing and sequencing protocols. Just as qRT-PCR workflows have been standardized over time (e.g., fixed cycle thresholds), sequencing-based workflows must also be calibrated to identify cost-effective and diagnostically meaningful depth. Further development of robust bioinformatics pipelines is also necessary to automate and improve the accuracy of viral detection. Our findings reinforce the need for tailored, population-specific viral reference databases for veterinary applications.\u003c/p\u003e\u003cp\u003eAlthough total RNA sequencing provides an unbiased method for virus detection, it requires higher sequencing depth and robust computational infrastructure. The decreasing costs of high-throughput sequencing have made metagenomics more accessible; however, data processing and storage remain significant barriers for implementation in large-scale or routine veterinary and agricultural diagnostics. Additionally, sample collection, preservation, and RNA extraction quality can strongly influence sensitivity. Targeted capture and enrichment methods have shown promise in improving viral detection sensitivity, but they are currently optimized for human diagnostics and are not widely adapted for veterinary pathogens (Buddle et al. \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eTogether, these findings support the potential of metatranscriptomics to enhance pathogen detection in veterinary diagnostics, particularly when conventional methods fall short of identifying novel pathogens. However, realizing this potential will require continued investment in reference curated database development, optimization of sequencing approaches, and bioinformatic tools for routine diagnostic analysis. By addressing these gaps, metatranscriptomic approaches can move closer to practical application in the field, offering a powerful complement to established molecular diagnostics.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to thank the eResearch Team from the Computational Research Support Unit (CRSU) at the University of Technology Sydney (UTS) for his invaluable support in providing the computational infrastructure and guidance for the bioinformatics analyses conducted in this study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatement of Animal Ethics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe cattle sampling procedure used in this study was approved by the University of Technology Sydney Animal Care and Ethics Committee (Approval No. ETH19-3407).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of Interest Statement\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFinancial support for this project was provided by UTS and the Australian Centre for Genomic Epidemiological Microbiology (AusGEM), a collaborative partnership between the NSW Department of Primary Industries and the University of Technology Sydney.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eGovender, K.N., Street, T.L., Sanderson, N.D. \u0026amp; Eyre, D.W. Metagenomic Sequencing as a Pathogen-Agnostic Clinical Diagnostic Tool for Infectious Diseases: a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies. \u003cem\u003eJ Clin Microbiol\u003c/em\u003e \u003cstrong\u003e59\u003c/strong\u003e,e0291620 (2021).\u003c/li\u003e\n \u003cli\u003eLiu, J., Zhang, Q., Dong, Y.Q., Yin, J. \u0026amp; Qiu, Y.Q. Diagnostic accuracy of metagenomic next-generation sequencing in diagnosing infectious diseases: a meta-analysis. \u003cem\u003eSci Rep\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e,21032 (2022).\u003c/li\u003e\n \u003cli\u003eNeyton, L.P.A., Langelier, C.R. \u0026amp; Calfee, C.S. Metagenomic Sequencing in the ICU for Precision Diagnosis of Critical Infectious Illnesses. \u003cem\u003eCrit Care\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e,90 (2023).\u003c/li\u003e\n \u003cli\u003eShakya, M., Lo, C.C. \u0026amp; Chain, P.S.G. Advances and Challenges in Metatranscriptomic Analysis. \u003cem\u003eFront Genet\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e,904 (2019).\u003c/li\u003e\n \u003cli\u003eCobbin, J.C., Charon, J., Harvey, E., Holmes, E.C. \u0026amp; Mahar, J.E. Current challenges to virus discovery by meta-transcriptomics. \u003cem\u003eCurr Opin Virol\u003c/em\u003e \u003cstrong\u003e51\u003c/strong\u003e,48-55 (2021).\u003c/li\u003e\n \u003cli\u003eBrito, B.P.\u003cem\u003e et al.\u003c/em\u003e Expanding the range of the respiratory infectome in Australian feedlot cattle with and without respiratory disease using metatranscriptomics. \u003cem\u003eMicrobiome\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e,158 (2023).\u003c/li\u003e\n \u003cli\u003eBushnell, B. BBMap: A Fast, Accurate, Splice-Aware Aligner. In: Institute, J.G., editor.; 2014.\u003c/li\u003e\n \u003cli\u003eMd, V., Misra, S., Li, H. \u0026amp; Aluru, S. Efficient Architecture-Aware Acceleration of BWA-MEM for Multicore Systems. \u003cem\u003eIEEE Parallel and Distributed Processing Symposium (IPDPS)\u003c/em\u003e; 2019.\u003c/li\u003e\n \u003cli\u003eDanecek, P.\u003cem\u003e et al.\u003c/em\u003e Twelve years of SAMtools and BCFtools. \u003cem\u003eGigascience\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e (2021).\u003c/li\u003e\n \u003cli\u003eConstantinides, B., Hunt, M. \u0026amp; Crook, D.W. Hostile: accurate decontamination of microbial host sequences. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e (2023).\u003c/li\u003e\n \u003cli\u003eWickham, H. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag; 2016.\u003c/li\u003e\n \u003cli\u003ePagès, H., Aboyoun, P., Gentleman, R. \u0026amp; DebRoy, S. Biostrings: Efficient manipulation of biological strings. R package version 2.74.1. 2024.\u003c/li\u003e\n \u003cli\u003eTeam, R.D.C. R: A language and environment for statistical computing. In: Computing, R.F.f.S., editor.; 2020.\u003c/li\u003e\n \u003cli\u003eParadis, E. \u0026amp; Schliep, K. ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e,526-528 (2019).\u003c/li\u003e\n \u003cli\u003eZhang, M.\u003cem\u003e et al.\u003c/em\u003e Assessment of Metagenomic Sequencing and qPCR for Detection of Influenza D Virus in Bovine Respiratory Tract Samples. \u003cem\u003eViruses\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e (2020).\u003c/li\u003e\n \u003cli\u003eCostello, M.\u003cem\u003e et al.\u003c/em\u003e Characterization and remediation of sample index swaps by non-redundant dual indexing on massively parallel sequencing platforms. \u003cem\u003eBMC Genomics\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e,332 (2018).\u003c/li\u003e\n \u003cli\u003eValiente-Mullor, C.\u003cem\u003e et al.\u003c/em\u003e One is not enough: On the effects of reference genome for the mapping and subsequent analyses of short-reads. \u003cem\u003ePLoS Comput Biol\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e,e1008678 (2021).\u003c/li\u003e\n \u003cli\u003eBuddle, S.\u003cem\u003e et al.\u003c/em\u003e Evaluating metagenomics and targeted approaches for diagnosis and surveillance of viruses. \u003cem\u003eGenome Med\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e,111 (2024).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"metagenomics, diagnostics, sensitivity, qRT-PCR, reference viral genome","lastPublishedDoi":"10.21203/rs.3.rs-7071637/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7071637/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn this study, the impact of sequencing depth and reference genome selection is shown to influence the detection of five bovine respiratory RNA viruses in metatranscriptomes from clinical cattle samples. Metatranscriptomes were generated from pooled samples and subsampled to 20\u0026nbsp;million, 10\u0026nbsp;million, and 1\u0026nbsp;million reads. We assessed the correlation between qRT-PCR cycle threshold (Ct) values and the number of reads mapped to reference genomes for four viruses: Bovine coronavirus (BCoV), Bovine nidovirus (BNV), Influenza D Virus (IDV), and Bovine Viral Diarrhea Virus-1 (BVDV-1).\u003c/p\u003e\u003cp\u003eStrong linear correlations were observed between mapped reads and Ct values. For BCoV and BNV, RefSeq genomes yielded detection thresholds at ~\u0026thinsp;Ct 38 with 1\u0026nbsp;million reads; IDV showed a threshold at Ct 34.4. Reference genome choice had minimal impact on BCoV, BNV, and IDV detection. However, BVDV-1 detection was poor using the divergent RefSeq genome and improved significantly with a sample-derived reference.\u003c/p\u003e\u003cp\u003eComplete coverage of the BCoV genome and IDV segment 6 was achieved in samples with Ct values below 30, regardless of the reference used. For BNV, genome 80% coverage was reached when using the NCBI RefSeq, even in samples with low-Ct samples. BVDV-1 could not be detected when using the RefSeq genome, highlighting the limitations of distant references.\u003c/p\u003e\u003cp\u003eThese findings demonstrate that both sequencing depth and reference genome choice substantially influence viral detection sensitivity in metatranscriptomic analyses.\u003c/p\u003e","manuscriptTitle":"Detecting RNA Viruses in Cattle: Effects of Sequencing Depth and Sequence References in Metatranscriptomics","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-21 18:05:56","doi":"10.21203/rs.3.rs-7071637/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"2882a98f-7fae-43e7-b44e-b52b58a2ea2d","owner":[],"postedDate":"July 21st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-08-02T22:38:15+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-21 18:05:56","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7071637","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7071637","identity":"rs-7071637","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}