3  Experimental Design for Viromics

A viromics project begins before the first command. In viromics, experimental design is more important than the pipeline — a good pipeline cannot rescue a poorly planned virome experiment. The research question controls the sample type, enrichment strategy, sequencing platform, controls, and analysis plan.

TipViromics is like fishing in a huge ocean

• Your sample type decides where you fish.
• Your enrichment method decides which net you use.
• Your sequencing platform decides how clearly you can see the fish.
• Your controls tell you whether you caught real fish or plastic that fell off your own boat.

Every design choice below is really one of these four decisions.

Learning objectives — by the end of this chapter you will be able to:

• turn a vague aim into a strong, testable viromics research question;
• choose a sample type, enrichment strategy, and DNA or RNA workflow for a given goal;
• explain how each enrichment step biases the observed community and how to report it;
• select a sequencing platform and the right set of negative and positive controls;
• plan biological versus technical replication for a defensible study; and
• build and validate a sample metadata sheet and FASTQ manifest on the command line.

3.1 Design workflow

Never start with the software. Start with the biological question and let it drive every downstream choice.

flowchart TD
  A[Research question] --> B[Sample type]
  B --> C[Enrichment strategy]
  C --> D[DNA or RNA extraction]
  D --> E[Controls]
  E --> F[Library preparation]
  F --> G[Sequencing platform]
  G --> H[Bioinformatics pipeline]
  H --> I[Biological interpretation]

flowchart TD
  A[Research question] --> B[Sample type]
  B --> C[Enrichment strategy]
  C --> D[DNA or RNA extraction]
  D --> E[Controls]
  E --> F[Library preparation]
  F --> G[Sequencing platform]
  G --> H[Bioinformatics pipeline]
  H --> I[Biological interpretation]

3.2 Start with a strong question

A weak question is too broad to design around:

What viruses are in my sample?

A strong question defines the environment, host, comparison, nucleic-acid type, and expected interpretation:

How does nitrogen fertilization change the DNA virome of maize
rhizosphere soil across crop growth stages?

The stronger version already tells you the sample type (soil), the host context (maize rhizosphere), the treatment (nitrogen fertilization), the comparison (across growth stages), and the analysis direction (differential DNA virome).

Field	Good viromics question
Human gut	How does antibiotic treatment affect gut bacteriophage diversity?
Soil	How do cropping systems change soil viral communities?
Plant pathology	Which RNA viruses are associated with symptomatic tomato leaves?
Wastewater	Can wastewater viromics detect enteric RNA viruses across seasons?
Animal surveillance	Which viruses are present in bat fecal samples from different habitats?

3.3 Sample types

Different sample types carry different viruses and different problems.

Sample type	Common viruses detected	Main challenge
Feces / gut	bacteriophages, enteric viruses	host and bacterial background, inhibitors
Soil	phages, plant viruses, fungal viruses	complex matrix, humic acids, low viral recovery
Water / wastewater	RNA and DNA viruses, phages	dilution, concentration required
Plant leaves	plant RNA viruses, phages, mycoviruses	plant RNA and DNA background
Blood / plasma	eukaryotic viruses, phages	very low biomass, contamination sensitive
Respiratory swabs	RNA viruses, bacteriophages	host RNA dominance
Animal tissues	host-associated viruses	host genome background

Low-biomass samples such as plasma, respiratory swabs, and some viral-like particle preparations are especially vulnerable to contamination, because even small amounts of reagent or environmental nucleic acid can become a large fraction of the final read set after amplification and sequencing. These samples need strict negative controls.

3.4 Enrichment methods

A virome sample contains far more non-viral than viral material — host cells, bacteria, fungi, archaea, free nucleic acids, and food, soil, or plant particles. Enrichment increases the proportion of viral particles before sequencing.

Method	Purpose	What it removes or enriches
Centrifugation	remove large debris and cells	plant tissue, soil particles, host cells
Filtration	remove cells larger than viruses	bacteria, eukaryotic cells
Nuclease treatment	degrade free DNA/RNA outside capsids	host and bacterial free nucleic acids
Viral concentration	increase viral particles	useful for water and wastewater
Density gradient	purify viral particles	selective but laborious
Random amplification	boost low-input viral nucleic acid	useful but may introduce bias

3.4.1 Enrichment is not neutral: what each step loses

Enrichment improves viral signal but changes the original community structure. Teach the trade-offs explicitly.

A 0.22 µm filter removes most bacteria, but may also lose:

• giant viruses,
• viruses attached to cells,
• viruses attached to particles, and
• fragile enveloped viruses.

Nuclease treatment reduces free host DNA and RNA, but may lose:

• damaged viral particles,
• unprotected viral genomes, and
• fragile RNA viruses.

Random amplification raises input from low-biomass samples, but may create:

• uneven genome coverage,
• overrepresentation of short templates,
• chimeric artifacts, and
• artificial abundance bias.

Warning

Enrichment improves viral signal, but it does not preserve the original community perfectly. Always report the enrichment method clearly so results can be interpreted and compared.

3.5 DNA virome or RNA virome

Choosing DNA or RNA is one of the most important design decisions, because it changes extraction, library preparation, and interpretation.

• A DNA virome targets dsDNA and ssDNA phages, archaeal and eukaryotic DNA viruses, and prophage-like sequences when total DNA is used.
• An RNA virome targets +ssRNA, -ssRNA, and dsRNA viruses, including most plant, animal, and clinically important viruses.

Feature	DNA virome	RNA virome
Starting molecule	DNA	RNA
Stability	more stable	less stable
Extraction	DNA extraction	RNA extraction
Reverse transcription	not required	required
Common targets	phages, DNA viruses	plant / animal / human RNA viruses
Main risk	host or bacterial DNA background	RNA degradation
Library prep	DNA library	cDNA library
Bioinformatics	DNA viral contigs	RNA viral contigs / transcripts

Use DNA viromics for phage ecology (soil, gut, marine), DNA viruses, viral dark matter, and prophages from microbial assemblies. Use RNA viromics for plant viral disease, respiratory and enteric viruses, zoonotic and wastewater RNA surveillance, and emerging-virus discovery. If you need both, design separate DNA and RNA workflows rather than forcing one library to do both jobs.

3.6 Total metagenome versus virus-enriched virome

Total metagenomes keep microbial context and prophages but can have very low viral read fractions, so viral assembly may be poor. Virus-enriched viromes improve viral recovery and low-abundance detection but lose cell-associated viruses, distort abundance, and can amplify contaminants. The right choice follows the question: host–virus association favours total metagenomes; viral community structure favours enrichment.

3.7 Sequencing platform

Illumina gives short, highly accurate reads and is the workhorse for community comparison and abundance. Nanopore gives long reads that can span whole viral genomes and enables rapid field work, at the cost of higher per-read error and less straightforward abundance estimation. Hybrid designs combine both: Nanopore for contiguity, Illumina for accuracy.

Goal	Recommended platform
Many samples, high accuracy	Illumina
Complete viral genomes	Nanopore or hybrid
RNA virus surveillance	Illumina or Nanopore
Low-input clinical virome	Illumina with strong controls
Rapid field detection	Nanopore
Publication-grade community comparison	Illumina, often with validation
Genome closure	Hybrid Illumina plus Nanopore

3.8 Controls

Controls are not optional; they are part of the bioinformatics interpretation. Clean negative controls should produce very few reads — many reads in a blank signal a major contamination risk.

Control	Purpose
Extraction blank	detect contamination from extraction kit and reagents
Library blank	detect contamination from library preparation
No-template control	detect PCR and amplification contamination
Mock viral community	check pipeline recovery and bias
Spike-in control	measure extraction and sequencing performance
Positive control	confirm the protocol can detect known viruses
Technical replicate	check sequencing and library reproducibility
Biological replicate	capture real biological variation

3.9 Common contamination sources

Viral metagenomics is especially sensitive to contamination, because unknown viral-like sequences may be misread as real biology when controls are missing.

Source	Example
Reagents	extraction kit nucleic acids
Lab environment	aerosols, previous PCR products
Cross-sample contamination	splashing, pipetting
Index hopping	reads assigned to the wrong barcode or index
Barcode cross-talk	multiplexing artifact
Host DNA / RNA	human, plant, or animal background
Bacterial DNA	incomplete filtration or lysis
PhiX	Illumina control spike-in
Adapter contamination	library-prep artifacts

3.10 Replication strategy

Biological replicates are different biological samples — soil from plot 1, plot 2, plot 3. They answer: is this pattern real in nature? Technical replicates are repeated processing or sequencing of the same sample — one DNA extract split into two library preps. They answer: is the method reproducible?

For student and small projects:

• Minimum: 3 biological replicates per group.
• Better: 5 or more biological replicates per group.
• Always: include negative controls.

3.10.1 Minimum teaching design

A defensible small project comparing two soil treatments:

Component	Count
Treatment A biological replicates	3
Treatment B biological replicates	3
Extraction blank	1
Library blank	1
Total libraries	8

That is 3 + 3 + 2 controls = 8 libraries, sequenced as Illumina paired-end 150 bp, with a mock viral community added if available.

3.11 A decision tree for planning

flowchart TD
  A[What is my target?] --> B{DNA or RNA viruses?}
  B -->|DNA viruses| C[DNA extraction] --> D[DNA virome pipeline]
  B -->|RNA viruses| E[RNA extraction] --> F[cDNA synthesis] --> G[RNA virome pipeline]
  B -->|Both| H[Run separate DNA and RNA workflows]
  D --> I{Is viral biomass low?}
  G --> I
  H --> I
  I -->|Yes| J[Add extraction and library blanks, spike-ins, contamination analysis]
  I -->|No| K[Standard controls]
  J --> L{Do I need complete genomes?}
  K --> L
  L -->|Yes| M[Nanopore or hybrid sequencing]
  L -->|No| N{Many samples and abundance comparison?}
  N -->|Yes| O[Illumina is usually practical]
  N -->|No| O

flowchart TD
  A[What is my target?] --> B{DNA or RNA viruses?}
  B -->|DNA viruses| C[DNA extraction] --> D[DNA virome pipeline]
  B -->|RNA viruses| E[RNA extraction] --> F[cDNA synthesis] --> G[RNA virome pipeline]
  B -->|Both| H[Run separate DNA and RNA workflows]
  D --> I{Is viral biomass low?}
  G --> I
  H --> I
  I -->|Yes| J[Add extraction and library blanks, spike-ins, contamination analysis]
  I -->|No| K[Standard controls]
  J --> L{Do I need complete genomes?}
  K --> L
  L -->|Yes| M[Nanopore or hybrid sequencing]
  L -->|No| N{Many samples and abundance comparison?}
  N -->|Yes| O[Illumina is usually practical]
  N -->|No| O

3.12 Metadata is part of the design

A viromics project without metadata is weak. Record at least the sample identity, environment, host, treatment, nucleic-acid type, enrichment, control type, platform, read type, and extraction and library batches. First set up the standard project folders you will reuse across the book.

⬇ setup_project.sh

# One-time: create the standard project structure used across the book
bash setup_project.sh          # or: bash ~/Downloads/setup_project.sh

3.12.1 Create the Phase 2 workspace

mkdir -p ~/viromics_course/phase2_experimental_design
cd ~/viromics_course/phase2_experimental_design
mkdir -p raw_reads metadata controls logs reports design_tables

3.12.2 Build the sample metadata sheet

cat > metadata/sample_metadata.tsv << 'EOF'
sample_id   environment host    treatment   replicate   nucleic_acid    enrichment  control_type    extraction_batch    library_batch   platform    read_type
Soil_A_01   soil    maize   Treatment_A 1   DNA filtration_nuclease biological_sample   batch1  lib1    Illumina    PE150
Soil_A_02   soil    maize   Treatment_A 2   DNA filtration_nuclease biological_sample   batch1  lib1    Illumina    PE150
Soil_A_03   soil    maize   Treatment_A 3   DNA filtration_nuclease biological_sample   batch1  lib1    Illumina    PE150
Soil_B_01   soil    maize   Treatment_B 1   DNA filtration_nuclease biological_sample   batch1  lib1    Illumina    PE150
Soil_B_02   soil    maize   Treatment_B 2   DNA filtration_nuclease biological_sample   batch1  lib1    Illumina    PE150
Soil_B_03   soil    maize   Treatment_B 3   DNA filtration_nuclease biological_sample   batch1  lib1    Illumina    PE150
EXT_BLANK_01    blank   none    none    NA  DNA none    extraction_blank    batch1  lib1    Illumina    PE150
LIB_BLANK_01    blank   none    none    NA  DNA none    library_blank   batch1  lib1    Illumina    PE150
EOF

column -t -s $'\t' metadata/sample_metadata.tsv

3.12.3 Create a FASTQ manifest

A manifest links each sample ID to its read files and prevents sample mix-ups.

cat > metadata/fastq_manifest.tsv << 'EOF'
sample_id   forward_read    reverse_read
Soil_A_01   raw_reads/Soil_A_01_R1.fastq.gz raw_reads/Soil_A_01_R2.fastq.gz
Soil_A_02   raw_reads/Soil_A_02_R1.fastq.gz raw_reads/Soil_A_02_R2.fastq.gz
Soil_A_03   raw_reads/Soil_A_03_R1.fastq.gz raw_reads/Soil_A_03_R2.fastq.gz
Soil_B_01   raw_reads/Soil_B_01_R1.fastq.gz raw_reads/Soil_B_01_R2.fastq.gz
Soil_B_02   raw_reads/Soil_B_02_R1.fastq.gz raw_reads/Soil_B_02_R2.fastq.gz
Soil_B_03   raw_reads/Soil_B_03_R1.fastq.gz raw_reads/Soil_B_03_R2.fastq.gz
EXT_BLANK_01    raw_reads/EXT_BLANK_01_R1.fastq.gz  raw_reads/EXT_BLANK_01_R2.fastq.gz
LIB_BLANK_01    raw_reads/LIB_BLANK_01_R1.fastq.gz  raw_reads/LIB_BLANK_01_R2.fastq.gz
EOF

3.12.4 Validate that sample IDs match

cut -f1 metadata/sample_metadata.tsv | tail -n +2 | sort > metadata/ids_metadata.txt
cut -f1 metadata/fastq_manifest.tsv | tail -n +2 | sort > metadata/ids_manifest.txt
comm -3 metadata/ids_metadata.txt metadata/ids_manifest.txt

comm -3 prints lines unique to either file. No output means the IDs match — a habit worth keeping before every run.

Estimated resources on 12 threads and 32 GB RAM: metadata validation finishes in seconds and uses under 1 MB of storage. When planning raw-data storage, start at 3x the expected compressed FASTQ size, because SRA download, decompression, and intermediate files can temporarily coexist.

WarningCommon mistakes

• Forgetting negative controls. In viromics, controls are part of the bioinformatics interpretation, not just wet-lab hygiene. No negative controls means weak contamination interpretation.
• Mixing DNA and RNA viromes without planning. They need different extraction, library prep, and interpretation — design them as separate workflows.
• Treating read abundance as true viral abundance. Read counts are shaped by genome length, extraction efficiency, amplification bias, library prep, sequencing depth, and enrichment. Use normalized measures such as coverage, RPKM, or TPM, and interpret carefully.

NoteExercise: design a plant RNA virome study

Create a metadata sheet for a wheat leaf RNA virome experiment with 3 healthy leaves, 3 diseased leaves, 1 extraction blank, and 1 positive control, all Illumina PE150 on an RNA workflow. Then count how many biological samples fall in each treatment group.

One solution:

mkdir -p ~/viromics_course/phase2_experimental_design/exercise1
cd ~/viromics_course/phase2_experimental_design/exercise1

cat > plant_rna_virome_metadata.tsv << 'EOF'
sample_id   environment host    treatment   replicate   nucleic_acid    enrichment  control_type    extraction_batch    library_batch   platform    read_type
Leaf_H_01   plant_leaf  wheat   healthy 1   RNA rRNA_depletion  biological_sample   batch1  lib1    Illumina    PE150
Leaf_H_02   plant_leaf  wheat   healthy 2   RNA rRNA_depletion  biological_sample   batch1  lib1    Illumina    PE150
Leaf_H_03   plant_leaf  wheat   healthy 3   RNA rRNA_depletion  biological_sample   batch1  lib1    Illumina    PE150
Leaf_D_01   plant_leaf  wheat   diseased    1   RNA rRNA_depletion  biological_sample   batch1  lib1    Illumina    PE150
Leaf_D_02   plant_leaf  wheat   diseased    2   RNA rRNA_depletion  biological_sample   batch1  lib1    Illumina    PE150
Leaf_D_03   plant_leaf  wheat   diseased    3   RNA rRNA_depletion  biological_sample   batch1  lib1    Illumina    PE150
EXT_BLANK_01    blank   none    none    NA  RNA none    extraction_blank    batch1  lib1    Illumina    PE150
POS_CTRL_01 mock    mock    mock    NA  RNA mock_community  positive_control    batch1  lib1    Illumina    PE150
EOF

# Count biological samples per treatment group
awk -F'\t' 'NR>1 && $8=="biological_sample" {count[$4]++} END {for (t in count) print t, count[t]}' \
  plant_rna_virome_metadata.tsv

Expected:

healthy 3
diseased 3

3.13 Key takeaways

• Experimental design matters more than the pipeline: a sharp, testable question — defining environment, host, comparison, and nucleic-acid type — should drive every downstream choice from sampling to analysis.
• Sample type, enrichment strategy, and the DNA-versus-RNA decision jointly determine which viruses you can detect; design separate DNA and RNA workflows rather than forcing one library to do both jobs.
• Every enrichment step (filtration, nuclease treatment, random amplification) biases the observed community, so the method must be reported clearly to make results interpretable and comparable.
• Negative and positive controls are part of the bioinformatics interpretation, not just wet-lab hygiene — especially for low-biomass samples where reagent contamination can dominate the reads.
• Biological replication (minimum three per group) answers whether a pattern is real, while a validated metadata sheet and FASTQ manifest keep samples traceable and the study reproducible (Minot et al. 2011).

3.14 Further reading

• Minot et al. (2011) demonstrates how sampling and enrichment choices shape an observed virome, motivating careful experimental design and controls.
• Chen et al. (2018) documents fastp, a fast all-in-one tool for the read quality control and adapter trimming that follows the sequencing choices planned here.
• The ICTV website provides the standard reference for naming and organizing the viruses your design aims to detect: https://ictv.global.
• The NCBI Sequence Read Archive hosts public raw reads useful for planning storage and benchmarking a design (NCBI 2026); browse it at https://www.ncbi.nlm.nih.gov/sra.

Minot, Samuel, Rohini Sinha, Jun Chen, Hongzhe Li, Sue A. Keilbaugh, Gary D. Wu, James D. Lewis, and Frederic D. Bushman. 2011. “The Human Gut Virome: Inter-Individual Variation and Dynamic Response to Diet.” Genome Research 21 (10): 1616–25. https://doi.org/10.1101/gr.122705.111.

Chen, Shifu, Yanqing Zhou, Yaru Chen, and Jia Gu. 2018. “Fastp: An Ultra-Fast All-in-One FASTQ Preprocessor.” Bioinformatics 34 (17): i884–90. https://doi.org/10.1093/bioinformatics/bty560.

NCBI. 2026. “Sequence Read Archive.” https://www.ncbi.nlm.nih.gov/sra.

3.15 Chapter figure

Figure 3.1: The viromics experimental-design workflow: a research question drives sample choice, enrichment, DNA or RNA extraction, controls, sequencing, and the bioinformatics pipeline.

Tip🎨 Generate this figure

Save as: images/ch02-experimental-design.png · Aspect ratio: 16:9 · Style: clean flat vector infographic, Codanics palette (teal #008b8b, navy #05043b, white background), no photorealism.

Prompt: Create a clean educational infographic showing the viromics experimental-design workflow as a left-to-right pipeline. Begin on the left with a labelled “Research question” card, then flow through labelled stages: “Sample type” (with small icons for soil, gut, wastewater, plant leaf, and respiratory swab), “Enrichment” (filter and nuclease icons), “DNA or RNA extraction” (two branching arrows, one DNA double helix, one RNA single strand), “Controls” (icons for extraction blank, library blank, and mock community), “Sequencing platform” (Illumina short-read and Nanopore long-read icons), and finally “Bioinformatics pipeline” shown as a Linux terminal leading to “Biological interpretation”. Add a subtle recurring fishing motif — a small net over the enrichment stage — to echo the fishing-in-the-ocean analogy. Use teal and navy Codanics branding, clear sans-serif labels, and connecting arrows between every stage.

3.16 Quiz: Experimental Design

Q1. Why are negative controls important in low-biomass viromics?

A. they remove the need for sequencing B. contamination can dominate low-biomass data C. they increase viral genome length D. they replace biological replicates

Note

Answer: B. Small contaminant signals can become a large fraction of the data when true biomass is low.

Q2. Which workflow requires reverse transcription?

A. DNA virome B. RNA virome C. protein annotation only D. Bowtie2 mapping

Note

Answer: B. RNA viromes are usually converted to cDNA before sequencing.

Q3. What is a disadvantage of viral enrichment?

A. it can introduce bias B. it prevents assembly C. it removes all viral reads D. it disables sequencing

Note

Answer: A. Filtration, nuclease treatment, and random amplification can change the observed community structure.

Q4. Which platform is usually practical for many high-accuracy paired-end samples?

A. Illumina B. Sanger only C. Western blot D. flow cytometry

Note

Answer: A. Illumina short reads are common for high-throughput community comparison.

Q5. What is the purpose of a FASTQ manifest?

A. link sample IDs to read files B. replace the reference database C. calculate Shannon diversity D. infer viral taxonomy

Note

Answer: A. A manifest prevents sample mix-ups and documents input files.

3.17 Interactive quiz: Experimental design

How to use this quiz: Select one option, click Check answer, and read the explanation. Use the reset button if you want to try again.

1. Which comparison is usually most fair when studying treatment effects in viromics?

A. One treated sample versus one control B. Biological replicates with consistent metadata and controls C. Only technical replicates D. No controls if sequencing depth is high

Biological replicates and good metadata are central to a defensible study design. Controls are also important for contamination tracking.

2. When is virus enrichment most useful?

A. When you want to increase the viral signal relative to host and microbial background B. When you want more ribosomal RNA C. When you want to avoid filtration D. When you only study proteins

Enrichment can improve viral signal by reducing non-viral background, though it may also introduce bias.

3. Why should extraction blanks be included?

A. To increase viral diversity B. To estimate genome size C. To detect contamination introduced during processing D. To replace biological replicates

Extraction blanks and negative controls help detect reagent and workflow contamination, which can be a major issue in low-biomass viromics.

4. What is a key difference between DNA and RNA virome workflows?

A. RNA viromes require reverse transcription before sequencing library preparation B. DNA viromes never need QC C. RNA viromes cannot be assembled D. DNA viromes do not need controls

RNA genomes need conversion to cDNA before sequencing in most common workflows.

5. Why is metadata design important?

A. It makes FASTQ files smaller B. It supports reproducibility and downstream analysis C. It replaces all statistics D. It avoids all contamination

Good metadata make samples interpretable and reusable. They also support reproducible analysis and clear reporting.