Meta-analysis of public raw sequence data unveils the distribution and dynamics of emerging aquatic pathogens: using Macrobrachium rosenbergii golda virus as a case study

TL;DR

This study uses a new database to analyze public sequencing data and reveals the global spread and biology of a prawn virus.

Contribution

The study introduces a novel method using the Logan database to efficiently mine SRA data for emerging aquatic pathogens.

Findings

MrGV is found in southern China, Thailand, and India, expanding its known geographic range.

MrGV has been circulating in Southern Asian prawn populations since at least 2011.

MrGV is predominantly associated with larval stages of Macrobrachium rosenbergii.

Abstract

Searching publicly archived sequence data for emerging aquatic animal pathogens is a powerful but challenging approach for increasing our understanding of newly identified or poorly characterized organisms. However, searching for target sequences within the sequence read archive (SRA) database requires significant time, data storage, and computing power, limiting its accessibility. Utilizing a new database, Logan, we undertook a meta-analysis of SRA data sets to investigate the presence of an emerging virus, Macrobrachium rosenbergii golda virus (MrGV). MrGV was first characterized in M. rosenbergii larvae in 2020, associated with repeated mass mortalities in Bangladesh hatcheries. MrGV has since been detected in two separate reports from the Jiangsu Province of central, coastal China, and during a larval mortality event in India. Here, we discovered that MrGV is present in two additional provinces in southern China, Thailand, and India. We also found molecular evidence to confirm, as previously suspected, the circulation of the virus within Southern Asian populations of M. rosenbergii as far back as 2011, and that, based on relative abundance, MrGV is mostly associated with larvae. Overall, the identification of MrGV sequences in data sets that are largely unpublished within the scientific literature has provided novel insights into the pathogen’s biology, including the prevalence of MrGV globally and the life stages of prawns that should be screened to prevent the spread of the virus. This work illustrates how mining public sequencing data, supported by databases like Logan and standardized metadata submissions, can support cost-effective epidemiological studies of pathogens and strengthen One Health approaches to global disease monitoring.

Searching for target sequences within the sequence read archive (SRA) database requires significant time, data storage, and computing power, limiting its accessibility. This study demonstrates how the Logan database, constructed from an SRA-wide genome assembly, can be utilized to rapidly and efficiently find target sequences within the SRA database, expanding the use of these publicly available data sets outside of their original intended purposes. Here, we searched for an emerging virus, Macrobrachium rosenbergii golda virus, in prawns to reveal insights into its geographic distribution, host range, and relative abundance, without the need for additional sampling. We demonstrate how, with careful application of this approach, alongside improvements in metadata quality and accessibility, sequencing data sets can be used to uncover critical insights into pathogen biology. This type of data mining could add otherwise unknown data to epidemiological studies of emerging, re-emerging, and rare pathogens globally, allowing the determination of the spread of agents within and between populations.

Figures

The presence of MrGV in M. rosenbergii SRA data sets by life stage. (A) Bar plot showing the number of M. rosenbergii SRA data sets containing MrGV sequences by life stage. Gray bars represent the absence of MrGV in Logan-assembled SRA contigs, and colored bars represent the presence of MrGV. (B) Violin plot showing the relative abundance of MrGV reads (reads/million, transformed with the inverse hyperbolic sine function) within M. rosenbergii SRA data sets published prior to 10 December 2023. SRA data sets with fewer than one MrGV read per million were excluded from the plot. PERMANOVA was used to compare relative abundance of MrGV between life stages of M. rosenbergii: * indicates P < 0.05 and * indicates P < 0.001.

Line plot showing the change in MrGV relative abundance over 20 days of larval development at different salinities (NCBI BioProjects PRJNA864119 and PRJNA891247). Salinity units from the SRA metadata were presumed to be in parts per thousand (‰). Each colored line represents development at a different salinity, with the 95% CI depicted by a ribbon in a lower transparency of the same color: blue represents development at 5‰, green represents development at 15‰, and purple represents development at 25‰.

Midpoint rooted Bayesian consensus tree constructed from 36 full genome sequences from MrGV. Circles on branches represent posterior probabilities of 1, and bold accession numbers represent genomes assembled in this study.

Midpoint rooted Bayesian consensus tree constructed from ORF3 nucleotide sequences (n = 48) from MrGV. Circles on branches represent posterior probabilities of 1, and bold accession numbers represent sequences generated in this study.

Full text

INTRODUCTION

The giant river prawn, Macrobrachium rosenbergii, is a key aquaculture species cultivated in tropical and sub-tropical regions around the globe, but it has suffered production issues due to multiple diseases (1). Macrobrachium rosenbergii golda virus (MrGV) was first characterized in larval stages of M. rosenbergii in 2020, associated with mortalities in multiple Southern Bangladesh prawn hatcheries (2). Two metatranscriptomic data sets of M. rosenbergii postlarvae and juveniles/adults from China have detected the presence of the virus; however, disease or poor condition was only noted in one animal (3, 4). During the course of this study, an investigation into large-scale mortality events in M. rosenbergii hatcheries in India during July to September 2024 found MrGV to be the causative agent (5). Despite these findings, the extent of MrGV presence in species of Macrobrachium globally is largely unknown.

MrGV falls within the Nidovirales order of enveloped positive-sense single-stranded RNA viruses. Nidoviruses have a wide host range, including both vertebrates and invertebrates (6). Invertebrate-infecting nidoviruses are a smaller group of viruses compared to those able to infect vertebrates (including SARS-CoV-2); however, their diversity is becoming better known through metatranscriptomic viral discovery studies (7, 8). Viruses within the Roniviridae family are some of the best recognized invertebrate-associated nidoviruses, comprising the yellow head viruses (YHV) (9), of which yellow head virus genotype 1 (YHV-1) is notifiable to the World Organisation for Animal Health (WOAH), due to its association with mass mortalities of penaeid shrimp (10, 11). Since its characterization, MrGV has been formally recognized and named by the International Committee of Taxonomy of Viruses (ICTV) as Nimanivirus lahi, within the newly erected second genus of Roniviridae, Nimanivirus (Subgenus: Marovirus) (12).

The sequence read archive (SRA) database has expanded almost exponentially as more studies produce short read metagenomic, transcriptomic, and metatranscriptomic data to answer specific research questions. Previously, searching the SRA database for a target sequence to address various research questions required huge resources in time, data storage, and computing power, making screening of SRA data sets largely inaccessible to most users. By utilizing massive cloud computing resources to carry out an SRA-wide genome assembly, the Logan data set of DNA and RNA sequences (13) has reduced the redundancy and data volume of the SRA data sets, allowing these data to be searched more efficiently and at an affordable cost to the user.

In this study, we performed a meta-analysis on the Macrobrachium spp. SRA database up to December 2023 for MrGV sequences, using Logan assembled SRAs to (i) further characterize the geographical spread of the virus, (ii) determine whether differences in MrGV nucleotide sequence are associated with geographical location using phylogenetics, and (iii) determine the species and life stages in which MrGV sequences are present, thereby demonstrating how the Logan database can be used to inform epidemiological studies of disease.

MATERIALS AND METHODS

Selecting SRA accessions and mapping Logan contigs

Logan assemblies of transcriptomes for all species of Macrobrachium deposited to the NCBI SRA database prior to 10 December 2023 were selected to be mined for MrGV sequences. The SRA accession numbers, their associated species, life stage, and country of origin are outlined and summarized in Tables S1 and S2. Logan-assembled contigs from release v1 were mapped to a MrGV reference genome (accession no. NC_076908) using minimap2 v2.28 with default parameters (14) within a custom Snakemake (15) v8.20.3 pipeline (logan-screen-accessions v1.0.0, DOI: 10.5281/zenodo.16418927). Samples were considered MrGV positive if they had at least 90% of bases covered by at least one contig, calculated using SAMtools v1.21 (16).

Relative abundance of MrGV reads

To calculate the relative abundance of MrGV within SRA data sets containing Logan-assembled contigs that mapped to the reference genome, raw reads from the MrGV-positive SRA data sets were mapped to the MrGV reference genome using minimap2 v2.28, as above. To compare abundance between SRA data sets, the presence of MrGV was calculated as MrGV reads per million and transformed using the inverse hyperbolic sine (asinh) function. Relative abundance plots were generated with the ggplot2 R package (17). Permutational Multivariate Analysis of Variance (PERMANOVA) was performed using the adonis2 function of Vegan v2.7-2 (18).

Construction of consensus MrGV genome sequences

Reads from biological replicates of SRA data sets with over 90% coverage of the reference genome were combined and mapped to the MrGV reference genome using minimap2, as described above, to increase coverage and enhance the likelihood of identifying small nucleotide polymorphisms (SNPs). The combined SRA data sets are detailed in Table S3. A consensus sequence from the mapped reads was generated using SAMtools v1.21 (16) in simple consensus mode, with call fraction set to 0.2. Combined raw reads from biological replicates (where they existed) were remapped to the resulting SAMtools-generated consensus sequence with minimap2, as above, and SNPs were called with Snippy v4.6.0 (19), with default settings. A consensus genome with IUPAC codes at variable sites within each SRA data set was generated from the raw output from snippy with the BCFtools v1.9 package (16), setting the quality of mapping to greater than or equal to 100, minimum coverage to 30, and the minimum proportion for a variant to be called to 0.1. For MrGV genomes already deposited to NCBI, the associated SRA sequences (where present) were downloaded, SNPs were called, and consensus sequences with IUPAC codes were generated, as above.

Phylogenetic analysis

A multiple sequence alignment (MSA) of complete MrGV genomes with degenerative nucleotide positions in IUPAC format and publicly available MrGV genomes was produced with MAFFT (20) using the L-ins-I algorithm. A Bayesian consensus tree was constructed from the MSA using MrBayes v3.2.7 (21) on the CIPRES Science Gateway (22). The tree was constructed using two separate MC^3^ runs, carried out for two million generations using one cold and three hot chains. The first 500,000 generations were discarded as burn-in, and trees were sampled every 1,000 generations. The resulting tree was midpoint rooted. A second Bayesian consensus tree, based on the ORF3 sequence, was generated as above.

RESULTS

MrGV presence in Macrobrachium spp. SRA data sets

A total of 965 Macrobrachium SRA data sets derived from RNA were screened for the presence of MrGV: 483 M. rosenbergii, 466 M. nipponense, 6 M. australiense, 4 M. cancinus, 2 M. novaehollandiae, 2 M. tolmerum, 1 M. olfersii, and 1 M. koombooloomba. MrGV was only found to be present in M. rosenbergii SRA data sets, appearing across all life stages, but with a notable preference for the larval life stage. Reads mapped to MrGV in 27% of M. rosenbergii embryo SRA data sets (n = 88), 61% of larval SRA data sets (n = 150), 41% of postlarvae SRA data sets (n = 41), and 7% of juvenile/adult SRA data sets (n = 201; Fig. 1A). MrGV was present at a significantly higher relative abundance in larvae compared to other life stages (Fig. 1B). Larvae had a median relative abundance of 6,013 MrGV reads/million, while embryos, postlarvae, and adults had median relative abundances of 3.73, 510.92, and 0.021 MrGV reads/million, respectively. Coverage across the MrGV genome was not uniform, with the majority of reads mapping to the 3′ end of the virus (Fig. S1).

The presence of MrGV in M. rosenbergii SRA data sets by life stage. (A) Bar plot showing the number of M. rosenbergii SRA data sets containing MrGV sequences by life stage. Gray bars represent the absence of MrGV in Logan-assembled SRA contigs, and colored bars represent the presence of MrGV. (B) Violin plot showing the relative abundance of MrGV reads (reads/million, transformed with the inverse hyperbolic sine function) within M. rosenbergii SRA data sets published prior to 10 December 2023. SRA data sets with fewer than one MrGV read per million were excluded from the plot. PERMANOVA was used to compare relative abundance of MrGV between life stages of M. rosenbergii: * indicates P < 0.05 and *** indicates P < 0.001.

The relative abundance of MrGV increases over larval development

SRA data sets that appear to relate to a timed developmental study on larval stages of M. rosenbergii at different salinities (BioProjects PRJNA864119 and PRJNA891247) showed an increase in MrGV read presence over the first 20 days of development. These BioProjects, which lack associated publications, sampled larvae in triplicate at 0-, 6-, 12-, 24-, and 48-hour post-hatch, followed by sampling on days 5, 10, and 20. Larvae were developed in salinities of 5, 15, and 25, with the units presumed to be parts per thousand (‰), although no units were stated in the SRA metadata. As there is no associated publication for these data, clinical signs of disease and mortality data were not available.

When transformed with the inverse hyperbolic sine function and plotted (Fig. 2), MrGV relative abundance follows the same pattern for larvae developed at each salinity, peaking at 10–15 days post-hatch, before plateauing. Average MrGV relative abundance stayed below 10 MrGV reads/million until 2 days post-hatch, where average MrGV reads increased to between 200 and 625 reads/million. Generally, the relative abundance MrGV reads continued to increase to approximately 7,000 reads/million between days 5 and 10, which was maintained until the end of the experiment at 20 days post-hatch.

Line plot showing the change in MrGV relative abundance over 20 days of larval development at different salinities (NCBI BioProjects PRJNA864119 and PRJNA891247). Salinity units from the SRA metadata were presumed to be in parts per thousand (‰). Each colored line represents development at a different salinity, with the 95% CI depicted by a ribbon in a lower transparency of the same color: blue represents development at 5‰, green represents development at 15‰, and purple represents development at 25‰.

MrGV was also present in M. rosenbergii embryos. Embryo SRA data sets with reads that mapped to MrGV originated from a single BioProject (PRJNA890910). This BioProject, submitted to NCBI by the same institute and at the same time as the larval salinity challenge, appears to study the transcriptome of M. rosenbergii over the course of embryonic development; however, no associated publication could be found, and we were unable to link the embryos to the salinity-challenged larvae. MrGV reads were present in all replicates (three per developmental stage) for all stages of embryonic development: first and second cleavage, blastula, gastrula, egg nauplius, egg metanauplius, egg protozoea, and egg zoea stages. However, the number of MrGV reads did not change over time and remained at a consistently low relative abundance, averaging between two and eight MrGV reads per million over the course of the study.

MrGV sequence type varies by location

A total of 88 SRA data sets were used to assemble 30 novel MrGV genomes. Reads from biological replicates within the same BioProject were combined in order to increase MrGV coverage and the ability to call SNPs. A further six MrGV genomes were obtained from NCBI GenBank. Of the 36 MrGV genomes, 31 originated from China, 2 from Thailand, 2 from India, and 1 from Bangladesh. The 31 genomes from China originated from three provinces: two neighboring provinces in the Southeast of China, Guangdong and Guangxi, and one province in the East of China, Jiangsu. None of the Logan contigs originating from M. rosenbergii SRA data sets from Israel, Malaysia, or Vietnam mapped to the reference MrGV genome.

A midpoint rooted Bayesian tree constructed from all complete MrGV genomes demonstrated that MrGV from each location branched separately from one another (Fig. 3). MrGV from Thailand, India, and Bangladesh branched as distinct, fully supported clades. MrGV from each province in China also branched separately, with full support for each clade. MrGV from the two geographically close provinces, Guangdong and Guangxi, branched as sister clades, while MrGV from the geographically distant province, Jiangsu, branched separately. Jiangsu is the only location where SRA data sets exist covering multiple years; despite this temporal difference, MrGV genomes still branch together. MrGV genome MW590703 had no associated SRA data set; therefore, the distance between this genome and other genomes from Jiangsu could be partially explained by the lack of degenerate bases within the genome, as variants could not be called.

Midpoint rooted Bayesian consensus tree constructed from 36 full genome sequences from MrGV. Circles on branches represent posterior probabilities of 1, and bold accession numbers represent genomes assembled in this study.

The nucleotide differences that cause MrGV from different locations to branch separately occurred across the MrGV genome, with the majority of these nucleotide differences (86.27%) not causing changes in the resulting amino acid sequence. The MrGV genome is organized into four major ORFs: ORF1a encodes a 3C-like protease, ORF1b encodes a replicase polyprotein hypothesized to be involved in viral RNA replication, ORF2 is a short hypothetical protein with no assigned function, and ORF3 is predicted to encode structural glycoproteins. Base changes that caused alterations in amino acid sequence accounted for 13.68%, 11.82%, 10.00%, and 21.18% of mutations in ORFs 1a, 1b, 2, and 3, respectively. Amino acid changes in the key nidovirus protein motifs within ORF1b were low, with no mutations that conferred an amino acid change in the S-adenosylmethionine-dependent N7-methyltransferases motif, and only one in the nidovirus RdRp-associated nucleotidyltransferase and zinc-binding domain motifs (Fig. S2). A large non-cytoplasmic domain at the 5′ end of ORF3, thought to encode one of the envelope glycoproteins, contained 14 sites in the 5′ region of the domain that resulted in a change in amino acid sequence. Given that the metadata associated with the SRA data sets were limited, we were unable to determine if these differences in glycoprotein sequence conferred a difference in the virulence or pathogenicity of MrGV. Variation in the 5′ untranslated region (UTR) of the MrGV genomes was not assessed due to low sequence coverage; however, the 3′ UTR, known to contain the presence of a secondary RNA structure nidoviruses (23), was well conserved between MrGV genomes from all locations, with only two nucleotide differences present (compared to the reference sequence) in MrGV from Guangxi, China.

A second Bayesian consensus tree was constructed from an MSA of the nucleotide sequence of ORF3 for all SRAs (or concatenated biological replicates) where a full ORF3 sequence was present. Due to non-uniform coverage across the MrGV genome, with the majority of reads mapping to the 3′ end of the virus (Fig. S1), ORF3 was chosen as it had the highest coverage and sequence availability for the most SRAs. The tree (Fig. 4) had more paraphyletic clades when compared to the tree constructed from full genomes (Fig. 3), and in general, had lower posterior probabilities. However, as in the tree generated from full genomes, samples from Guangdong and Guangxi branched separately from other MrGV samples with maximal support, and within this clade, they branched as highly supported sister clades. Samples from Jiangsu also formed a separate clade, but with lower support (0.82 posterior probability) than in the tree constructed from full genomes.

Midpoint rooted Bayesian consensus tree constructed from ORF3 nucleotide sequences (n = 48) from MrGV. Circles on branches represent posterior probabilities of 1, and bold accession numbers represent sequences generated in this study.

DISCUSSION

In this study, we found MrGV present in M. rosenbergii SRA data sets from multiple locations in Asia. Previously, MrGV had only been reported from Bangladesh (2), one province in China (Jiangsu) (3, 4), and India (5). However, we found MrGV sequence data in prawns from Thailand and in an additional two provinces in the southeast of China; we also found MrGV in SRAs from India from 2011. In addition to the geographical distribution of MrGV, we determined that the presence and high relative abundance of MrGV is highly associated with the larval life stage of M. rosenbergii and is apparently absent from other species within the Macrobrachium genus, although SRA data sets for larval stages of other Macrobrachium species were limited (Table S2). Despite adult and juvenile SRAs making up the largest proportion of M. rosenbergii data sets, very few animals had Logan contigs that mapped to the MrGV genome, and MrGV relative abundance was low.

Mass mortalities of M. rosenbergii larvae in Bangladesh hatcheries first occurred in 2011 (2). Although no samples of moribund larvae exist to confirm MrGV infection in these early mortalities, the clinical signs of disease were identical to MrGV infection. Here, we show that MrGV was also present in larvae and postlarvae (SRA accession nos. DRR023219 and DRR023253, respectively) from Odisha, India, at the same time that the mortalities started to occur in Bangladesh; however, the metadata associated with the larvae and postlarvae from India state that these animals were healthy. Interestingly, the study that describes MrGV as the causative agent of mass mortalities in M. rosenbergii hatcheries in India in 2024 (5) was from the same institute in India that submitted the SRA datasets in 2011 that we found to contain MrGV. Future investigation should be directed toward understanding how MrGV was introduced into hatcheries in both Bangladesh and India in 2011 and determining why, if MrGV has been present in Odisha since 2011, mortalities have only started occurring recently. Particular attention should be given to understanding whether M. rosenbergii genetics or rearing conditions play a role in the susceptibility of larvae to MrGV.

The study that initially characterized MrGV suggested that additional work was required to determine how the virus was entering hatchery systems (2). In this study, we find MrGV reads in SRA data sets for all stages of M. rosenbergii embryo development, suggesting that vertical transfer of MrGV is a possibility. Paul et al. (5) also found MrGV to be present at low abundance in M. rosenbergii eggs collected from adults; therefore, collection of wild and/or unscreened berried female M. rosenbergii as broodstock for hatcheries may be a route of transfer of MrGV into hatcheries. Stress has long been portrayed as a major facilitator in the progression of disease in aquaculture (24), and the stress of culture conditions may be a factor in the susceptibility of larvae to MrGV infection. M. rosenbergii larvae perform best at a salinity of 13‰ (25); hence, the salinity stress delivered by culturing M. rosenbergii larvae at salinities higher or lower than this value, for example, those used in BioProjects PRJNA864119 and PRJNA891247 (Fig. 2) may predispose animals to infection with MrGV.

The changes in relative abundance of MrGV over larval development provide knowledge of the point in larval development when MrGV becomes abundant within prawn larvae. Due to the pooling of larvae from multiple time points by Hooper et al. (2), it was only possible to determine that MrGV was lower in abundance 1–4 days post-challenge than MrGV to 5–10 days post-challenge. A similar trend was observed in the salinity-challenged larvae (Fig. 2), with the initial increase in MrGV relative abundance associated with larvae 2 days post-hatch, and MrGV relative abundance peaking between days 5 and 10 post-hatch before plateauing. The increase in MrGV relative abundance coincided with the time in larval development during which M. rosenbergii larvae are typically within the zoea III stage (26). Paul et al. (5) report that larval mortalities occur in hatchery settings in India from zoea III to zoea IX. As the timed larval study (Fig. 2) only demonstrates the relative abundance of MrGV up to 20 days post-hatch, it misses the later zoea stages; however, MrGV abundance was highest when larvae were likely in the zoea III–VII stages.

This study was limited by substantial data gaps and inconsistencies within the metadata associated with the SRA database. One of the most significant challenges was the ability to link publications to data from BioProject, BioSample, or SRA identifiers to fill these metadata gaps. Publication and data set identifiers are not automatically connected by NCBI, and identifiers are often inconsistently cited or omitted from publications. Given these missing data, particularly those associated with the health status, clinical signs of disease or tissue that was sequenced, we were unable to associate the presence of M. rosenbergii with poor health or moribundity. The current study was also limited by the number of larval data sets available in the SRA database. As the Logan database only included data that were available in the SRA prior to the end of 2023, no assembled SRA sequences were available for larval stages of Macrobrachium species except for M. rosenbergii, and SRA data sets for postlarvae were limited to 45 data sets from M. nipponense and one from M. australiense. Given these limitations, we cannot confidently determine that the host range of MrGV is limited to M. rosenbergii, assuming that if it is able to infect other species of Macrobrachium, it would be primarily associated with the larval life stage. To improve the utility of public sequencing data for pathogen surveillance, we recommend standardizing metadata submissions to include host, location, developmental stage, and environmental context.

A lack of SRA metadata regarding the health status of the host meant that it was not possible to associate any amino acid changes in key MrGV motifs to the infectivity of MrGV. Mutations in glycoprotein motifs in other nidoviruses have been shown to affect functional properties, including altering viral infectivity. Due to the large number of genomes and extensive patient data associated with infection with SARS-CoV-2, single mutations that conferred amino acid changes within the spike glycoprotein could be associated with increased or decreased binding to host receptors and subsequently linked to a change in infectivity (reviewed in reference 27). To a lesser extent, this type of study has been carried out within the YHV complex of viruses, where only genotypes 1 and 2 have been associated with yellow head disease, while other genotypes persist at low levels in healthy animals (28). It has been suggested that variation in a critical cleavage site within the YHV glycoprotein gp116 regulates the availability of viral glycoproteins, ultimately influencing whether the virus is able to cause chronic persistent infection or acute infection and disease (28). Conversely, even with a large deletion within the gp116 cleavage site in YHV-1, the virus remains highly virulent and infectious, despite significant structural deformation that reduces its incorporation into virions (29). Given this, information on the health status of animals associated with viral detection is essential to link predictions of virulence inferred from sequence changes to the ability of a pathogen to cause disease.

Summary

This study demonstrates how the Logan database can be utilized to rapidly and efficiently find target sequences within the SRA database, expanding the use of these publicly available sequencing data sets outside of their original intended purposes, for example, using transcriptomic studies to find associated pathogens. Here, we used the Logan database to search for an emerging virus (MrGV) in prawns and demonstrated how this can add valuable knowledge about the virus, including geographical spread, host range, and relative abundance, without the need for additional sampling or experimental infection. However, metadata associated with SRA data sets are typically inconsistently entered and poorly maintained. Without consistent, comprehensive metadata, the reliable interpretation of sequencing data remains challenging. Careful application of this approach, alongside improvements in metadata quality and accessibility, could uncover critical insights into pathogen biology, transmission, and control. This study illustrates how mining public sequencing data can support cost-effective pathogen surveillance and strengthen One Health approaches to global disease monitoring. This type of investigation could add otherwise unknown data to epidemiological studies of emerging, reemerging, and rare pathogens globally, allowing the determination of the spread of agents within and between populations and govern which compartments of a system are most appropriate to screen to prevent pathogen spread.