Soil virome analysis uses sequencing and computational methods to investigate viruses present in soil communities, including viruses that may infect bacteria, archaea, fungi, plants and other hosts. If you need to read soil viral operational taxonomic units (vOTUs), CheckV genome completeness, putative viral contigs, proviruses, viral read coverage or predicted virus–host links, first identify which viral sequences were recovered and how their identities were established.
A viral contig is not automatically a complete infectious virus, a vOTU is not necessarily an officially named viral species, and a predicted bacterial host is not proof that infection occurred in the sampled soil. A CheckV quality score addresses a particular property of an assembled viral sequence, not whether the virus was active or abundant in absolute physical units. Soil virome interpretation therefore requires a chain of evidence, from sample collection and DNA recovery through sequence identification, genome quality, clustering, mapping and host inference.
This educational guide explains that chain in plain modern scientific English. It separates directly sequenced DNA from inferred viral origin, model-estimated completeness from physical infectivity, and read-supported detection from observed ecological effects. Original arithmetic examples show how to check vOTU clustering conditions, breadth of read mapping, length-adjusted representation and contamination claims. Illustrative figures are teaching examples, not measurements of any real field or laboratory experiment.
Quick answer: what does a soil virome table contain?
A soil virome report may list viral candidate contigs, groups of related viral sequences called vOTUs, quality and completeness estimates, mapped sequencing reads, functional gene predictions and putative cellular hosts. Each output describes a different aspect of an inferred viral community. A sequence table shows candidate genetic material, while a coverage matrix gives evidence of reads matching those candidates in samples.
None of these fields by itself establishes how many active viruses were present per gram of soil. A proposed host association says which organism a computational or experimental test connects with a viral candidate; it does not always show an infection in the original field. Read the row and column definitions before drawing ecological conclusions. The same colourful heatmap could represent raw reads, length-adjusted coverage or relative abundance, and those units are not interchangeable.
Why soil viruses deserve their own analysis
Microbial viruses can influence host mortality, gene exchange and microbial food-web dynamics, although the extent and direction of each influence must be measured in context. In soil, viruses encounter heterogeneous pores, minerals, moisture conditions and host communities. Some are free particles and others are integrated into host genomes or associated with cells. A bulk DNA sample can contain several of these forms.
That complexity creates a distinctive reader question: how does a study move from fragmented environmental DNA to statements about viral populations? A report should preserve the distinctions between free viruses, intracellular viral DNA and prophages. Sampling, extraction and sequencing choices influence which forms are recoverable. A meaningful interpretation recognises those choices rather than assuming the instrument captured every virus inhabiting the soil.
Begin with the sample preparation route
A virome-enriched preparation may filter or concentrate virus-like particles before sequencing, while an unfractionated metagenome may sequence DNA from cells and environmental material together. The two approaches sample different fractions of the ecosystem. Virus-like-particle enrichment can improve representation of some extracellular viruses but may miss integrated or cell-associated forms. Bulk metagenomics may recover proviruses along with cellular genomes.
Soil complicates extraction because particles can adsorb nucleic acids or viruses, and different soils respond differently to processing. Ask whether results came from particle enrichment, total DNA, metatranscriptomics or another method. When comparing two studies, describe the sampled fraction before comparing viral diversity. The laboratory route sets the evidence boundary for the entire downstream analysis.
Soil mass and extraction recovery are not automatic denominators
A plot showing many viral reads does not report viral particles per gram simply because soil was weighed during collection. DNA recovery, particle elution, filtration, amplification and sequencing can vary across samples. A count of sequence matches is a technical measurement conditioned on those operations. It may not correspond linearly to the number of intact viruses in the original material.
To estimate physical abundance, researchers would need a validated quantitative method, its calibration, detection limits and sample-mass normalisation. The viral metagenomic sequence table may remain very useful without that conversion. A careful caption names relative sequence representation rather than silently replacing reads with particles. This protects the difference between the amount sequenced and the amount in nature.
Viral discovery is a classification task
After sequencing and assembly, computational programmes may identify contigs carrying viral hallmark genes, virus-like gene composition or other evidence. No single signature reliably recognises every unknown virus. Some cellular elements resemble viruses, some viruses lack familiar hallmark genes and short sequences can contain too little information for confident classification.
Ask which discovery method was used, its version, the sequence-length filters and how doubtful candidates were handled. A file called viral_contigs.fna may contain predictions, not universally verified viral genomes. Examine method confidence categories, alternative classifications and any manual checks before reporting a new viral lineage. Classification confidence is separate from completeness: a strongly viral-looking fragment can still be incomplete.
What a contig is and what it is not
A contig is a contiguous sequence assembled or otherwise represented from reads. Its length is usually expressed in nucleotides or base pairs. It may correspond to part of a viral genome, a nearly complete genome, a proviral region or an erroneous reconstruction. Assembly can fail across repeats, variable regions or low-coverage segments; it can also produce chimeras under difficult conditions.
Do not call every contig a complete virus or a virus isolate. A contig may have hundreds of genes or only a small fragment, depending on the organism and sequencing method. The reported nucleotide length establishes how much sequence was recovered, not how much of the expected viral genome it represents. Completeness requires additional evidence or a justified model.
Why viral genome completeness is difficult
Bacterial metagenome-assembled genomes are often assessed with sets of conserved single-copy marker genes. Viruses are too varied for one universal conserved marker set to play the same role across all groups. Viral genome length, gene content, segmentation and replication strategies differ markedly. That is why viral quality estimation uses methods tailored to viral biology.
A completeness estimate may come from comparison with reference viral genomes, viral protein families or structural evidence of a closed sequence. For a lineage distant from available references, a programme may be unable to estimate completeness reliably. ‘Undetermined’ is therefore a scientifically meaningful result, not necessarily an analytical failure. It marks a limit on the available evidence.
What CheckV actually does
CheckV is a computational tool developed to assess the quality of metagenome-assembled viral sequences. It can identify and trim host-derived sequence around integrated proviruses, estimate completeness where suitable references permit, and examine evidence for complete or closed genome structures. Its integrated summary classifies sequences into quality tiers.
The original CheckV study explains the method and its limits. The official CheckV source provides version-specific implementation information. CheckV does not confirm infection, viability or ecological process rates. Treat it as a genome-sequence quality instrument. It is not a substitute for establishing viral identity with appropriate evidence in the first place.
Read the quality summary before the colour legend
The CheckV quality_summary.tsv table can include contig identifiers, lengths, predicted completeness, completeness method, a quality category, inferred provirus status, contamination estimates and warnings. Do not interpret a single percentage without the method and caveats in neighbouring columns. A high-looking estimate may be flagged or based on limited reference support.
Inspect whether the sequence is complete, high-quality, medium-quality, low-quality or of undetermined completeness under the software’s rules. Record the software and database versions because those affect reference matching. If a paper reports only a grouped bar chart of quality tiers, look for the underlying table before drawing a conclusion about the reliability of specific sequences.
A high-quality category is not the same as biological activity
CheckV commonly places sequences estimated above 90% completeness in a high-quality tier; lower tiers cover less complete fragments, while undetermined means completeness could not be estimated. A predicted closed genome is a separate structural assessment. These classifications concern sequence reconstruction under the software’s reference and terminal-repeat checks.
A high-quality viral genome does not show that an infectious particle was recovered. A complete sequence might be present as DNA in a host or environmental sample, and a low-quality fragment might still originate from an active virus. Genome quality is necessary for some analyses but does not establish the timing, place or mode of viral infection. Keep quality and activity in separate sentences.
Host contamination has a specific meaning here
In viral sequence quality control, host contamination can refer to flanking cellular DNA included when a prophage or provirus region was extracted from a larger host-derived contig. CheckV identifies potential boundaries and can trim those regions for downstream viral analysis. This is not identical to laboratory sample contamination or to the completeness and contamination metrics used for bacterial MAGs.
A provirus is viral sequence integrated into a host genome or represented in an integrated state. Removing host flanks can prevent cellular genes from being mislabelled as viral functions. The presence of a host flank may also contain useful evidence about the host association, if the assembly is credible. Report the original and trimmed context appropriately rather than treating every trimmed base as proof of a dirty laboratory.
Worked trimming example with units
Imagine an illustrative assembled sequence of 48,000 base pairs. A boundary method estimates that the viral region occupies 40,000 base pairs and the remaining 8,000 base pairs are host-associated flanks. The fraction of the original contig outside the estimated viral region is 8,000 divided by 48,000, or about 16.7%.
That number is a fraction of sequence length flagged as nonviral in this invented example. It is not a statement that 16.7% of the soil sample was contaminated, nor that 16.7% of sequencing reads were erroneous. It depends on boundary accuracy. A report should state which sequence and operation the percentage describes. The denominator changes the biological meaning.
Terminal repeats can suggest closed genome structure
Direct terminal repeats, inverted terminal repeats and flanking host regions can provide clues that an assembled viral sequence reaches the boundaries expected under some replication or integration strategies. However, repeat sequences may also arise through assembly artefacts or repeated genomic elements. A software flag for terminal repeats is evidence to inspect, not an automatic guarantee of a complete biological genome.
Check the estimated length, assembly support and software warnings. Some viral genomes are segmented, so one complete segment is not the whole genome. Others use unusual structures or are poorly represented in the reference collection. Good wording is ‘putatively complete under the reported structural and computational criteria’ when further experimental confirmation has not been provided.
What MIUViG reporting standards contribute
The Minimum Information about an Uncultivated Virus Genome, or MIUViG, framework sets out recommended reporting information for environmental viral genomes, including origin, identification, genome quality, annotation, taxonomy, biogeography and host predictions. It helps readers judge how much of a sequence is known and what evidence underlies its biological labels.
MIUViG is not itself a sequencing instrument or a way to infer infectivity. Its value is transparent metadata and consistent descriptions. The MIUViG standards paper distinguishes fragments, high-quality drafts and finished genomes with explicit expectations. Avoid confusing a quality-tier name with official validation of every ecological claim made later in the article.
Why a vOTU is an operational cluster
A viral operational taxonomic unit groups sequences according to declared similarity and alignment-coverage criteria. It allows analysts to reduce redundant genomes and compare related viral populations across samples. The cluster is operational because its identity depends on the computational definition and reference sequences used. It need not coincide exactly with an officially recognised taxonomic species.
Many studies use high average nucleotide identity thresholds around 95% alongside an alignment-coverage requirement, but the exact alignment denominator and algorithm must be stated rather than assumed. A cluster created under one rule may differ from one created under another. A vOTU count therefore means ‘clusters meeting these criteria’, not a census of all biological virus species in the soil.
Average nucleotide identity needs aligned sequence context
Average nucleotide identity, or ANI, describes sequence identity over aligned regions under a given comparison method. A 95% ANI result says that roughly 95% of compared aligned nucleotide positions agree under that calculation. It does not necessarily mean two genomes are 95% identical over their entire lengths, because only a subset may align.
Imagine one sequence is 20,000 bases and another 40,000. A 95% identity over just 2,000 aligned bases provides much less evidence of overall similarity than a high-identity alignment spanning nearly all of the shorter sequence. This is why vOTU clustering usually has a separate coverage requirement. Report both the percent identity and the breadth definition.
Worked example: ANI and alignment breadth
Suppose viral candidate A is 30,000 bases long and candidate B is 28,000. Their aligned region covers 26,000 bases of B. Alignment breadth against the shorter sequence is 26,000 divided by 28,000, approximately 92.9%. If 25,000 of those aligned bases match, simple identity over that region is 25,000 divided by 26,000, approximately 96.15%.
Under an illustrative rule requiring at least 95% identity and at least 85% of the shorter sequence aligned, this pair passes both thresholds. Those numbers teach the calculations; real aligners may apply additional filters and definitions. Do not infer an official species name solely from passing these thresholds. Also distinguish alignment breadth from read coverage when reading downstream sample profiles.
Why vOTU cluster counts depend on fragmentation
A fragmented viral genome might be represented by two contigs that fail to overlap enough for clustering even though both originated from one biological genome. Conversely, conserved genome regions may cause unrelated or more distant sequences to appear similar locally. Clustering rules, assembly completeness and reference collection affect the number of reported units.
If treatment A has more vOTUs than treatment B, ask whether sequencing depth, assembly quality, sample number and quality thresholds were comparable. A larger cluster count does not automatically establish greater underlying viral richness. Use carefully normalised or sensitivity-tested summaries when comparing groups. Operational units are a useful common language, but they remain dependent on how sequences were recovered.
Read mapping supplies a different kind of evidence
After candidate genomes or vOTUs are defined, researchers may map reads from individual soil samples back to those reference sequences. The number and locations of matching reads provide evidence that related sequence material was present in the libraries. This is different from predicting the source sequences themselves and from establishing that intact viruses or hosts were active.
Mapping parameters matter: sequence identity, minimum alignment length, handling of multimapping reads and breadth of reference coverage can all affect detection. If one highly conserved viral gene attracts reads from several relatives, those reads may not distinguish which genome was actually present. A report should describe how ambiguous assignments and uneven coverage were handled.
Coverage depth and breadth are different measures
Coverage depth describes the number of mapped bases over reference positions on average or at specified sites. Breadth describes the fraction of the reference sequence with at least some qualifying mapping coverage. A small highly conserved region can show deep coverage while the rest of the genome has none. High depth over one gene is not equivalent to widespread support across a viral genome.
When reading a vOTU presence matrix, identify any minimum breadth criterion. Ask whether the declared genome was supported across its length or merely through a short region. Mapping depth can also vary with library size and other biases. Both dimensions help distinguish robust detection from matches that might arise because multiple viral genomes share a conserved segment.
Worked breadth calculation
Imagine a 40,000-base reference viral contig. Qualifying reads cover 30,000 distinct base positions at least once. The observed breadth of coverage is 30,000 divided by 40,000, or 0.75, meaning 75% of positions are covered under the chosen mapping rules. If the same reads had concentrated on only 4,000 positions, breadth would be just 10%.
These calculations do not establish genome completeness. They describe how much of an already defined reference is supported by mapped reads in a particular sample. An incomplete reference can have 100% mapping breadth. Never translate read breadth of 75% as ‘the viral genome is 75% complete’ unless the study separately established that very different claim.
Normalising read counts by length and library size
Longer reference genomes offer more positions for reads to align, and larger sequencing libraries contain more reads to allocate. Some analyses therefore report length-adjusted and library-adjusted counts, using measures such as reads per kilobase per million mapped reads, or RPKM. The exact denominator depends on the protocol and software output.
A valid comparison names the reference length, number of assigned reads and total reads used for normalisation. This transformed value is still a measure of sequence representation, not viral particles per gram. Different sample-processing routes and mapping decisions can make direct cross-study comparisons unreliable even when both tables use the same familiar acronym.
Worked RPKM example
Suppose a 30-kilobase viral reference receives 120 qualifying reads in a library where the selected normalisation denominator is 2 million mapped reads. The simplified RPKM calculation is 120 divided by 30 kilobases and then divided by 2 million-read units. This gives 2 reads per kilobase per million mapped reads.
If the number of mapped reads is incorrectly supplied as 2 rather than 2 million-read units in a formula already converting raw counts, the answer can be wrong by a factor of a million. Make units explicit. The result does not reveal an absolute virus count. It is a scaled library mapping statistic whose interpretation depends on reliable reference detection.
Viral abundance is not the same as viral diversity
A sample may be dominated by a few highly represented vOTUs, while another distributes reads among many less abundant units. Richness counts how many units pass detection criteria, while evenness describes the distribution of observed representation. Neither is simply the same as total sequence read count. Differences in assembly and mapping affect both.
Use consistent quality and coverage filtering when comparing viral communities. The soil microbial alpha-diversity guide explains general richness and evenness concepts, but viral vOTU data have their own preparation and detection biases. A useful caption identifies whether it describes genome clusters, aligned reads or ecological populations.
Predicted viral hosts are usually hypotheses
Computational host predictions may use CRISPR spacer matches, sequence composition, homology with known viruses, shared genome signatures or integration evidence. These methods provide different forms of support. A high-confidence sequence match under a reliable criterion may strengthen a candidate host link, but a predicted relationship does not necessarily show infection of that host in the sampled soil.
Ask what taxonomic rank is assigned—strain, species, genus or family—and whether the method provides a calibrated confidence measure. A host predicted at family level should not be rewritten as a specific named bacterial strain. If the hypothesis affects an important ecological conclusion, seek multiple independent lines of evidence or direct infection experiments where feasible.
CRISPR matches need their own interpretation
Some bacteria and archaea carry CRISPR spacers derived from exposure to foreign genetic material. A strong match between a spacer and a viral sequence can suggest a historical interaction or exposure with that host lineage. However, spacers can be shared among close relatives and match quality matters. The precise bacterial host and timing of interaction may remain uncertain.
A host prediction based on CRISPR evidence should document the spacer source, sequence identity, alignment span and database context. Avoid the leap from historical genetic evidence to current active infection in the same soil sample. The soil microbial genome quality guide is relevant when host evidence derives from bacterial or archaeal assembled genomes.
Proviruses can supply strong contextual clues
When a viral region is confidently identified within a cellular genome contig, flanking sequence may link the virus to a host lineage. This can be stronger contextual evidence than a broad k-mer similarity score, provided the assembly is not chimeric and the cellular taxonomic assignment is trustworthy. It also highlights why trimming and retaining original assembly context serve different purposes.
The viral sequence is separated for some comparative analyses, while the verified host-flanking context may be preserved as host-link evidence. Neither implies that the integrated element was induced, replicated or released as an infectious particle in the original sample. State whether the result concerns an integrated genome state or an observed free-virus fraction.
A predicted phage is not a verified pathogen
Viruses that infect microbes are commonly called bacteriophages when the hosts are bacteria, but soil virome datasets may contain viral sequences from multiple host domains. Even within bacterial viruses, infection outcomes differ. Some integrate into host genomes, some replicate lytically and many are poorly characterised. Broad gene signatures do not always reveal the complete life cycle.
A paper should not claim that a predicted phage killed a specific bacterial population without suitable temporal, laboratory or experimental evidence. A host link indicates possible biological association; host mortality requires independent support. Distinguishing those claims is particularly important when discussing crop health, soil fertility or putative biological control applications.
Viral auxiliary metabolic genes require careful checking
Some viruses encode genes associated with host metabolism, often discussed as auxiliary metabolic genes. Identification can be confounded by nearby host genes, imperfect assemblies and ambiguous gene annotations. A pathway-like annotation on a candidate viral contig does not by itself prove that the virus alters nutrient cycling in soil.
Verify that the gene lies within a credible viral region, that the annotation is specific enough and that the genome context supports viral origin. Removing host flanks can materially affect interpretation. Even a well-supported viral gene is potential capacity until expression or process effects are shown. Report ‘viral-encoded candidate function’ unless the ecological role has been experimentally established.
Why viral and bacterial MAG quality cannot be substituted
A bacterial MAG completeness estimate often concerns recovery of expected marker genes, with contamination assessed from duplicated or discordant markers. Viral genome quality uses different assumptions and evidence because no universal viral single-copy marker set exists. The terms high-quality and completeness may appear in both studies, but their operational definitions and validation differ.
Avoid presenting a CheckV percentage as if it were an ordinary bacterial CheckM-style score. Read the specific quality guidelines for the data type. Cross-kingdom comparisons of predicted ‘complete genomes’ may be unfair when assembled sequence lengths, reference resources and marker availability differ. The correct unit and method belong next to every percentage.
Watch for circular inference in genome classification
A viral candidate may be discovered with one algorithm, scored for completeness against related references and given a host prediction using overlapping sequence databases. These stages are not always independent pieces of evidence. If all rely on the same small cluster of reference genomes, agreement among tools may partly reflect shared input information.
A robust explanation distinguishes genuinely independent validation from repeated reuse of the same sequence relationships. When methods disagree, inspect why rather than choosing the most appealing label. Evidence provenance matters: a confident-looking consensus is less persuasive when every tool depends on the same unrepresentative database.
Contamination controls matter beyond CheckV
Negative extraction controls, sequencing blanks and laboratory processing records can reveal contaminants introduced during DNA preparation. This laboratory contamination is conceptually different from the cellular flanks around a provirus addressed by CheckV. Both can create misleading viral candidates but require different checks.
When a rare vOTU appears only in samples processed on one day, investigate batch effects and control libraries. A viral-looking contig may be technically real sequence material but originate outside the intended soil environment. Do not hide such possibilities behind a completeness score. Genome quality and sample provenance are separate requirements for reliable interpretation.
Sampling depth and season can dominate comparisons
Soil moisture, depth, plant-associated microhabitats and sampling time can affect viral and host communities. A difference between amended and untreated soils may be confounded if one group was sampled deeper or later. Biological replicates are independent plots or appropriately defined field units, not merely multiple libraries from the same extract.
Show the study design, sample sizes and variance before attributing vOTU changes to a treatment. A heatmap is not a causal experiment. Even if all viral genomes were perfectly assembled, an observational comparison could still be confounded by soil management history and physical conditions. Sequence certainty does not create experimental independence.
Comparative virome claims need consistent pipelines
Two studies may use different viral-enrichment techniques, sequencing technologies, discovery tools, contig cutoffs, completeness databases and mapping thresholds. Their raw counts of viral contigs are therefore not directly equivalent. One may identify many short fragments while another retains only high-quality genomes, creating an apparent diversity difference largely due to filtering.
A useful comparison reanalyses data under compatible methods where possible, or explicitly reports why equivalence cannot be assumed. If one paper’s category is ‘high-quality draft’ and another uses ‘CheckV high quality’, inspect definitions before merging summaries. Methodological compatibility is a prerequisite for biological comparison, not a minor technical footnote.
A misleading soil health headline and its repair
Imagine a report claiming, ‘Treated soil has twice as many harmful viruses because viral read counts doubled.’ The figure shows total mapped reads after sequencing, but the treatment group had deeper libraries and a higher fraction of virus-like sequence calls. No viral infectivity, host mortality or harmful ecological outcome was measured. The claim confuses technical representation with biological danger.
A corrected sentence might read: ‘The treatment libraries contained more reads assigned to the specified viral reference set under the study’s mapping method; further normalisation, host evidence and experimental work are needed to evaluate ecological consequences.’ The statement preserves the observable pattern and identifies the missing causal evidence.
What a credible validation study could do
A stronger investigation could sequence replicated soil treatments over time, apply consistent DNA processing, retain extraction controls and map reads under predetermined quality rules. Candidate host relationships could be checked through independent genomic signals, host-range experiments or other appropriate methods. Viral activity could be investigated with measurements designed for the relevant virus–host system.
The experimental design should match the biological question. Validating a particular viral genome sequence is not the same as demonstrating infection or a nutrient-cycling effect. A carefully staged validation plan separates identification, genome quality, host association and ecological consequence. That sequence makes claims narrower but considerably more useful.
Technical vocabulary: virus, contig, genome and population
‘Virus’ can refer to an infectious entity or a broader biological lineage depending on context; ‘viral sequence’ refers to nucleic acid evidence. ‘Contig’ is an assembled sequence; ‘genome’ is a complete genetic complement when supported. ‘vOTU’ is an operational sequence cluster and ‘host’ is an organism that may support infection, not merely any taxon co-occurring in a sample.
Words such as complete, detected, abundant, pathogenic and active are often stronger than their underlying measurements. The cross-linguistic transfer guide helps readers preserve technical distinctions, while the translation and meaning guide explains why context can change how a familiar word should be read.
A practical ten-minute virome reading route
Start at the sampling method: enriched particles or bulk metagenome? Next find the viral detection and assembly software, the minimum contig length and the reference database. Inspect CheckV quality alongside warnings, then identify the vOTU clustering rules. Read mapping depth, breadth and scaling before interpreting the abundance matrix. Finally inspect how hosts and functions were predicted.
Write two sentences about every major claim. One should name what was measured or computationally inferred. The other should name what would be needed to demonstrate the ecological conclusion. If a figure cannot support that separation, ask for the original methods and results. Clear scientific reading is a sequence of evidence checks, not a contest to remember the most acronyms.
Frequently asked questions about soil viromes
Does a vOTU equal an official viral species? Not necessarily. Does CheckV directly test infectivity? No. Can a high-quality contig contain a viral genome that was inactive at sampling? Yes. Does 100% read-mapping breadth mean the reference is a complete viral genome? No; those are distinct measurements. Can host DNA flank a provirus? Yes, and boundaries matter.
Do more reads always mean more viral particles per gram? No. Can an unknown viral fragment still be genuine? Yes, with appropriate identification evidence. Does a CRISPR match prove present-day infection? No. Are all soil viruses bacterial phages? No. Can viral gene annotation alone prove a nutrient-cycle effect? No. Each answer becomes more useful when paired with the relevant sampling or validation detail.
Independent practice with explained answers
A viral reference is 50,000 bases long and mapped reads cover 35,000 unique positions. What is breadth? Thirty-five thousand divided by fifty thousand equals 70%. Does this mean its genome is 70% complete? No; coverage of a reference and completeness relative to an expected biological genome are different. Another contig contains a predicted 45,000-base viral region within a 60,000-base assembly. What fraction lies outside the region? Fifteen thousand divided by sixty thousand equals 25%.
Two illustrative contigs align for 18,000 bases, with 17,100 matching. The simple aligned identity is 95%, but one contig is 40,000 bases long. Can you cluster them automatically using a rule requiring at least 85% alignment of the shorter sequence? Not without knowing the length of the shorter contig and how the rule measures breadth. Identity alone is insufficient.
Sources and relevant eduKate learning routes
Use the primary CheckV genome-quality study, the official CheckV project and the MIUViG reporting guidelines to examine the exact definitions of completeness, viral genome evidence and quality reporting. These sources should be read alongside the version and reference database used in the soil study being interpreted.
For distinct neighbouring questions, see eduKateSG’s guides to soil shotgun functional-gene tables, microbial metagenome-assembled genomes and microbial co-occurrence networks. They use related data but answer different reader questions.
The lasting lesson: classify the evidence before the virus
Soil virome sequencing can reveal previously unknown viral genetic diversity and generate precise hypotheses about microbial hosts and ecological roles. Its strength comes from linking sequence discovery to transparent genome quality, reproducible operational clustering, defensible read mapping and appropriately qualified host predictions. Its weakness appears when a model label is promoted into direct evidence of infection or ecosystem impact.
The most useful translation preserves every boundary: candidate viral sequence versus verified particle, contig versus complete genome, vOTU versus named species, mapped reads versus physical counts, predicted host versus demonstrated infection, and functional gene potential versus measured process. Once those distinctions remain visible, a complex virome report becomes an understandable and testable scientific explanation.