Unbiased Plant Virus Discovery by Total RNA Sequencing: rRNA Depletion, Sequencing Depth, and Reporting

Ribosomal RNA-depleted total RNA sequencing can survey polyadenylated and non-polyadenylated RNA viruses, viroids, viral transcripts, and mixed infections without requiring a primer for every expected agent. It is most informative when tissue captures the biological signal, extraction preserves RNA, depletion chemistry fits the host, sequencing depth is tied to a stated evidence goal, and analysis combines reference mapping with de novo assembly and translated similarity searches. A viral read or contig is discovery evidence, not automatic proof of an active infection, causal relationship, host assignment, complete genome, or absence from samples in which it was not detected. Independent confirmation and transparent reporting therefore belong in the original design.

Key takeaways

• Define whether the project seeks known viruses, divergent relatives, complete genomes, coinfections, or exploratory virome evidence.
• Pair symptomatic and relevant asymptomatic material while controlling tissue, stage, environment, and batch.
• Choose total RNA and depletion chemistry according to the viral genome space and plant organelle background.
• Set depth from expected host background, pooling, viral abundance, and required genome recovery rather than a universal read count.
• Evaluate candidates with read, contig, coverage, complexity, taxonomy, and control evidence together.
• Confirm priority candidates from an independent extraction and separate discovery from causality or risk conclusions.

Start With the Discovery Question

Known-target PCR and unbiased sequencing answer different questions. PCR is efficient when the candidate list and genomic targets are defined. Total RNA high-throughput sequencing is valuable when symptoms are unexplained, several viruses may coexist, variants could escape primers, or novel agents are plausible. "Unbiased" describes the absence of a virus-specific enrichment target; the experiment still contains biases from tissue selection, extraction, depletion, library construction, sequencing, database coverage, and analysis thresholds.

The project brief should identify the primary evidence endpoint. Detection-level evidence may accept a confirmed partial region. Strain characterization may require broad genome coverage. Novel-virus work may require genome completion, terminal confirmation, taxonomic placement, and follow-up host studies. Combining every goal under one generic "deep sequencing" request makes depth and reporting impossible to evaluate.

Sample the Biological Signal

Plant viruses are not distributed uniformly. Titer can vary among leaf, petiole, phloem, meristem, root, fruit, seed, developmental stage, cultivar, temperature, and time since infection. Collect tissue where the suspected agent is biologically likely to accumulate, but do not sample only the most necrotic area; degraded tissue can contain poor RNA and secondary organisms. A composite across several young symptomatic leaves may improve representation when it preserves the defined biological unit.

Include comparisons that allow interpretation. Symptomatic and visually unaffected plants should be matched for cultivar, age, location, tissue, and collection time where possible. For a single affected plant, collect spatially separated tissues and retain a second aliquot for independent extraction. Field blanks or collection controls can reveal handling contamination, while healthy host controls help establish background sequences and index cross-talk.

Record at minimum:

• stable sample ID, host species, cultivar or accession, organ, tissue position, and developmental stage;
• symptom description, onset, distribution, severity, and photographs;
• field or growth conditions, treatment history, location, collection date, and collector;
• pooling rule, number of plants per pool, tissue mass contribution, and reserve aliquot;
• stabilization method, interval to freezing, transport temperature, storage, and freeze-thaw events;
• extraction batch, library batch, index combination, and every negative or positive control.

Pooling saves libraries but changes the inference. A positive pool does not identify the contributing plant, and dilution may suppress a low-titer agent. If plant-level prevalence or follow-up is important, retain individual tissues and use a staged design: discover in carefully defined pools, then test individuals with targeted confirmation.

Choose the RNA Fraction

Total RNA with plant rRNA depletion is a broad strategy for RNA virus and viroid discovery because it does not depend on a poly(A) tail. Standard poly(A) selection enriches host messenger RNA and polyadenylated viral RNAs but may underrepresent or miss non-polyadenylated genomes and viroids. Small RNA sequencing captures host-derived virus small interfering RNAs and can support broad detection, yet its short reads create different assembly and classification challenges. Double-stranded RNA or virion-associated nucleic acid enrichment can improve particular signals while narrowing the recovered molecular population.

The correct fraction depends on the target space:

• use rRNA-depleted total RNA for a broad survey of RNA viruses, viroids, and viral transcripts;
• consider small RNA when antiviral small-RNA evidence is central or total RNA quality is limiting;
• consider poly(A) selection only when its exclusion of non-polyadenylated targets is acceptable;
• add a DNA-oriented library when circular or other DNA viruses are within scope and RNA evidence alone may be insufficient;
• preserve untreated nucleic acid or tissue so an alternative fraction can be tested later.

Pecman and colleagues compared total-RNA approaches across Illumina and nanopore platforms and showed that library preparation affects sensitivity and genome recovery. The small RNA sequencing service is an alternative when a project is designed around virus-derived small RNAs rather than conventional total-RNA libraries. The direct RNA sequencing resource provides context for native RNA and long-read applications, but direct RNA sequencing and short-read depleted total RNA should not be treated as interchangeable.

Treat Depletion as Experimental Design

In plant RNA, cytosolic, chloroplast, and mitochondrial rRNAs can consume a large fraction of reads. A kit described only as "rRNA depletion" may target human, bacterial, or a limited set of plant sequences. Confirm compatibility with the host species or close relatives, and determine whether cytoplasmic and organellar rRNAs are covered. Polyploid crops, divergent wild relatives, algae-associated samples, and poorly characterized hosts may show incomplete depletion.

Evaluate depletion with more than final library yield. Report RNA input, integrity or fragment distribution, depletion chemistry and version, rRNA fraction before and after processing when measurable, library size distribution, duplicate behavior, and fractions mapping to host nuclear and organellar references. If the host genome is incomplete, use rRNA databases and taxonomic classification rather than relying on one reference assembly.

Controls should include a negative host matrix and, when appropriate, a noncompetitive process control or RNA spike that can monitor extraction, depletion, library construction, and sequencing. A process control does not replace a biological positive and must not be so abundant that it increases cross-sample contamination. Randomize biological groups across extraction and library batches and use unique dual indexes where available.

Set Depth by Evidence Goal

There is no universal read count that guarantees plant virus discovery. Viral abundance can differ by orders of magnitude, and useful reads are the fraction remaining after quality filtering, rRNA depletion, host removal, duplication, and control review. Depth should be planned around expected viral titer, host complexity, number of pooled plants, genome size, desired coverage, coinfection complexity, and whether divergent discovery depends on assembly.

Use pilot rarefaction when uncertainty is high. Reanalyze progressively larger read subsets and plot candidate detection, genome breadth, contig length, and assembly stability. A plateau suggests more sequencing may add little for that sample; a continuing rise can justify deeper data. If depletion failed, resequencing the same poor-complexity library may be less effective than rebuilding it.

The agricultural NGS services page describes broader platform options, while the transcriptome RNA-seq service is relevant to depleted total-RNA library construction. Neither a platform name nor a nominal gigabase total replaces a project-specific usable-read and coverage plan.

Separate Host From Virus

Analysis should preserve evidence at each transition. Begin with read-quality review, adapter and low-quality trimming, duplicate assessment where appropriate, and control-aware contamination checks. Remove or classify host, chloroplast, mitochondrial, and rRNA reads while retaining counts so depletion and background can be audited. An incomplete host reference can leave plant-derived reads that resemble distant viral proteins, so screening should not depend on host subtraction alone.

Combine reference mapping with de novo assembly. Mapping is sensitive for known or moderately divergent viruses but can miss novel regions. Assembly can create longer interpretable candidates but is affected by depth, repeats, coinfections, chimeras, and parameter choice. Search reads and contigs against nucleotide and protein databases. Translated searches are often critical for divergent RNA viruses whose nucleotide similarity has eroded.

Record database names and dates, host assembly and annotation versions, software versions, key parameters, taxonomy rules, and thresholds for retaining a candidate. The agricultural transcriptomic data analysis service can support RNA-seq data processing, but plant-virus discovery requires additional virology-specific reference review, control interpretation, and candidate classification beyond routine differential expression.

Evaluate Viral Candidates

Rank evidence across independent dimensions. A credible candidate may have several non-overlapping reads, a supported contig, coherent open reading frames, recognizable conserved domains, plausible genome organization, consistent read orientation, coverage across the genome, absence from negative controls, and reproducibility in related biological samples. No single metric is universally decisive.

Review these questions for every candidate:

• Do reads occur in separate genomic regions or only one conserved motif?
• Are alignments unique, complex, and consistent with the claimed taxon?
• Does the contig have read support across junctions and ends?
• Is coverage broad or concentrated in low-complexity or host-like regions?
• Are similar reads present in blanks, unrelated libraries, or neighboring indexes?
• Could the sequence derive from an integrated element, reagent contaminant, fungus, insect, or other organism associated with the plant?
• Does genome architecture support a known virus family or a coherent novel lineage?

Candidates should retain a status such as detected, provisional, confirmed, unresolved, contaminant-like, or rejected, with the rule that produced it. A ranked spreadsheet without these rules cannot show whether two analysts would reach the same conclusion. Preserve borderline candidates in a review appendix when they fail only one threshold, especially when an alternative sample or deeper library could resolve them.

Mixed infections require agent-specific summaries. Do not infer relative viral load directly from raw read counts because genome size, RNA form, replication strategy, depletion, and mapping rules differ. Use normalized values as evidence descriptors, not universal titers. The NGS in animal and plant breeding resource gives broader agricultural NGS context; virus-candidate interpretation needs its own taxonomic and contamination safeguards.

Confirm Independent Evidence

Priority candidates should be confirmed with a complementary assay designed after the sequence is reviewed. Use a new extraction from retained tissue or an independently collected sample when possible. Target at least one informative region, and for a novel or high-consequence candidate, confirm separated genome regions, assembly junctions, and uncertain termini. Sanger sequencing of the confirmation product can verify identity rather than merely producing a band of the expected size.

Confirmation has distinct levels:

• technical confirmation shows that the candidate sequence is reproducible outside the original library;
• biological confirmation shows occurrence in independent plant material;
• host-association evidence separates plant infection from an associated organism;
• causal evidence links the agent to symptoms through epidemiology, localization, transmission, or experimental work;
• taxonomic characterization supports naming and placement under current criteria.

Fontdevila Pareta and colleagues emphasize that newly discovered sequences require staged characterization and risk analysis. A verified viral sequence should not be described as the cause of disease solely because it came from a symptomatic plant, especially in a coinfection. Similarly, failure to confirm a low-abundance candidate may reflect sampling heterogeneity and should be reported rather than erased.

Report What Was Not Found

"Not detected" is conditional on the submitted tissue, nucleic-acid fraction, controls, sequencing yield, depletion performance, analysis database, pipeline, and reporting threshold. It is not proof that a plant, field, seed lot, or species is virus-free. State the usable read count, residual rRNA and host fractions, control behavior, minimum candidate rule, and any sample-specific limitation.

Distinguish failure states:

• insufficient or degraded RNA before library construction;
• library failure or low complexity;
• excessive rRNA or host background;
• inadequate depth for the stated goal;
• no retained viral candidate under the defined analysis;
• candidate detected but not independently confirmed;
• partial sequence insufficient for strain or taxonomic assignment.

Transparent negative reporting prevents later users from treating technical failure as biological absence. It also allows a rational decision among deeper sequencing, a rebuilt library, another tissue, a different RNA fraction, or targeted follow-up.

Specify the Analysis Package

A reusable delivery package should include the sample and batch manifest, raw FASTQ files, read-QC summaries, depletion and host-classification metrics, filtered reads where agreed, candidate read and contig FASTA files, mapping files or coverage tables, de novo assemblies, nucleotide and protein search results, taxonomy evidence, control comparison, consensus sequences with ambiguity codes, and a candidate decision table.

For each candidate, report provisional name, closest references, accession numbers, aligned lengths, identities, expected genome organization, recovered breadth, depth distribution, contig support, control status, confirmation status, and interpretation limit. Provide machine-readable tables beside figures. The outsourcing transcriptome analysis guide offers general delivery-planning context, and the plant abiotic stress RNA-seq design guide reinforces the value of biological replication and batch balance even though virus discovery has different endpoints.

Plant Virus Sequencing Support

CD Genomics supports agricultural plant-virus research with consultation on tissue and control design, total RNA intake, host-compatible rRNA depletion, RNA library construction, short-read sequencing, and custom bioinformatics for reference mapping, de novo assembly, translated similarity search, candidate review, and documented delivery. Projects can include predefined decision gates for additional depth, alternative RNA fractions, genome-gap closure, or confirmation-assay design. The service scope is for research and does not provide clinical diagnosis, patient testing, treatment decisions, or medical claims.

Plant Virus Discovery FAQ

References

1. European and Mediterranean Plant Protection Organization. PM 7/151 (1) Considerations for the use of high throughput sequencing in plant health diagnostics. EPPO Bulletin. 2022;52(3):619–642.
2. Massart S, Adams I, Al Rwahnih M, Baeyen S, Bilodeau GJ, et al. Guidelines for the reliable use of high throughput sequencing technologies to detect plant pathogens and pests. Peer Community Journal. 2022;2:e62.
3. Pecman A, Adams I, Gutiérrez-Aguirre I, Fox A, Boonham N, Ravnikar M, Kutnjak D. Systematic Comparison of Nanopore and Illumina Sequencing for the Detection of Plant Viruses and Viroids Using Total RNA Sequencing Approach. Frontiers in Microbiology. 2022;13:883921.
4. Fontdevila Pareta N, Khalili M, Maachi A, Rivarez MPS, Rollin J, et al. Managing the deluge of newly discovered plant viruses and viroids: an optimized scientific and regulatory framework for their characterization and risk analysis. Frontiers in Microbiology. 2023;14:1181562.
5. Maina S, Donovan NJ, Plett K, Bogema D, Rodoni BC. High-throughput sequencing for plant virology diagnostics and its potential in plant health certification. Frontiers in Horticulture. 2024;3:1388028.
6. Kanapiya A, Amanbayeva U, Tulegenova Z, Abash A, Zhangazin S, Dyussembayev K, Mukiyanova G. Recent advances and challenges in plant viral diagnostics. Frontiers in Plant Science. 2024;15:1451790.
7. González-Pérez E, Chiquito-Almanza E, Villalobos-Reyes S, Canul-Ku J, Anaya-López JL. Diagnosis and Characterization of Plant Viruses Using HTS to Support Virus Management and Tomato Breeding. Viruses. 2024;16(6):888.
8. Villamor DEV, Ho T, Al Rwahnih M, Martin RR, Tzanetakis IE. High Throughput Sequencing For Plant Virus Detection and Discovery. Phytopathology. 2019;109(5):716–725.

This content and the described services are intended for agricultural and biological research. They do not provide clinical diagnosis, treatment decisions, or individual health assessment.