- Whole-genome sequencing (WGS) is a comprehensive DNA sequencing approach used to determine the nucleotide sequence of an organism’s entire genome. Unlike targeted sequencing, which examines selected genes or genomic regions, whole-genome sequencing aims to analyse coding and noncoding regions across the nuclear genome and, where relevant, organellar genomes such as mitochondrial and chloroplast DNA. It provides a broad view of genetic variation, genome organization, gene content, structural changes, and evolutionary relationships. WGS has become an essential method in bioinformatics, genomics, molecular biology, medicine, agriculture, microbiology, population genetics, and evolutionary research.
- The central principle of whole-genome sequencing is to extract genomic DNA from a biological sample, prepare it for sequencing, generate millions or billions of sequence reads, and use computational methods to reconstruct or analyse the genome. Depending on the sequencing technology, DNA molecules may be fragmented into shorter pieces or preserved as long molecules. The resulting reads are subjected to quality control, alignment or de novo assembly, and downstream analysis. Bioinformatics pipelines transform raw sequencing data into biologically meaningful information, including single-nucleotide variants, insertions, deletions, structural variants, copy-number changes, and other genomic features.
- Whole-genome sequencing differs from whole-exome sequencing (WES), targeted gene sequencing, and other specialized sequencing approaches. WES primarily examines protein-coding regions of the genome, which represent only a small proportion of the total human genome. Targeted sequencing focuses on predefined genes or genomic regions associated with a particular disease, trait, or biological process. In contrast, WGS examines coding regions, introns, intergenic regions, regulatory sequences, repetitive DNA, and many other genomic features. This broader coverage allows researchers to investigate variants outside protein-coding regions, although interpreting their biological significance can be challenging.
- Whole-genome sequencing can be performed using different sequencing technologies. Short-read sequencing platforms, particularly Illumina systems, generate large numbers of relatively short DNA reads, commonly ranging from approximately 50 to 300 base pairs. These reads generally provide high per-base accuracy and are widely used for germline and somatic variant analysis. Long-read sequencing technologies, including PacBio HiFi and Oxford Nanopore sequencing, generate much longer DNA reads. These can span repetitive regions, structural rearrangements, and complex genomic sequences that are difficult to resolve using short reads. The choice of technology depends on the research objective, genome complexity, required accuracy, read length, sequencing depth, throughput, and available resources.
- The whole-genome sequencing workflow begins with the collection of a biological sample. The sample may consist of blood, saliva, tissue, cultured cells, microbial material, plant tissue, or another source of biological DNA. Sample selection depends on the organism and research question. For human germline sequencing, blood or saliva is commonly used, while tumour tissue may be used for cancer genome analysis. In microbial genomics, DNA may be extracted from a cultured isolate or directly from a mixed environmental or clinical sample. Proper sample identification, collection, storage, and contamination control are essential for reliable results.
- DNA extraction is performed to isolate genomic DNA from the biological material. The extraction procedure generally involves cell lysis, removal of proteins and other contaminants, purification of nucleic acids, and recovery of DNA. DNA concentration can be measured using fluorometric methods, while purity can be assessed using absorbance-based measurements. DNA integrity and fragment-size distribution are also important, particularly for long-read sequencing. Contamination with RNA, proteins, salts, or other substances can interfere with library preparation and sequencing performance. For ultra-long-read applications, extraction procedures must minimize mechanical damage and preserve high-molecular-weight DNA.
- After extraction, DNA quality control is performed to determine whether the sample is suitable for sequencing. Important measurements include DNA concentration, purity, integrity, and fragment length. Fluorometric assays such as Qubit are commonly used for DNA quantification, while spectrophotometric measurements can provide information about purity. Gel electrophoresis and automated fragment-analysis instruments can help evaluate DNA integrity and size distribution. The requirements depend on the sequencing platform and library preparation method. A sample suitable for short-read sequencing may not be appropriate for an ultra-long-read experiment.
- The next step is library preparation. A sequencing library is a collection of DNA molecules modified with specific adapters and other components required by the sequencing platform. For short-read sequencing, genomic DNA is usually fragmented into smaller pieces using mechanical or enzymatic methods. Sequencing adapters are attached to the fragments, and sample-specific index sequences may be added to enable multiplexing. In some workflows, unique molecular identifiers (UMIs) are incorporated to distinguish original DNA molecules from PCR duplicates or to support error correction. The prepared library is then quantified, assessed, and loaded onto the sequencing platform.
- For PacBio sequencing, library preparation commonly involves the creation of circular SMRTbell templates. These templates allow DNA polymerase to pass repeatedly around the original DNA insert, enabling circular consensus sequencing and the generation of highly accurate HiFi reads. Oxford Nanopore library preparation involves attaching sequencing adapters and motor proteins to DNA molecules, allowing them to pass through nanopores. Long-read and ultra-long-read workflows require particular attention to DNA integrity because the original molecule length influences the maximum achievable read length.
- Sequencing is performed using an appropriate instrument and sequencing chemistry. Short-read platforms generate large numbers of relatively short reads, while long-read platforms can sequence much longer individual molecules. Sequencing may be performed in single-end or paired-end mode for short-read technologies. Paired-end sequencing reads both ends of a DNA fragment and provides additional information about fragment orientation and approximate insert size. Long-read technologies generally produce continuous reads from individual DNA molecules, providing extended sequence context.
- Sequencing depth, also known as coverage, is a major consideration in whole-genome sequencing. It describes how many times, on average, each position in the genome is represented by sequencing reads. Coverage is commonly expressed as a multiple of the genome size, such as 10×, 30×, or 60×. For example, 30× coverage means that the total number of sequenced bases is approximately thirty times the size of the genome, although actual coverage varies across genomic positions. Higher coverage can improve the detection of variants, particularly when sequencing quality is high and reads are distributed relatively evenly. However, increasing coverage also increases sequencing costs and computational requirements.
- The theoretical average sequencing coverage can be estimated using the following relationship:
Coverage = (Number of reads × Average read length) / Genome size
- For paired-end sequencing, the total number of sequenced bases from both reads should be considered. For example, if 100 million read pairs are generated and each read is 150 bases long, the total number of sequenced bases is:
Total bases = 100,000,000 × 2 × 150
= 30,000,000,000 bases
- For a genome of approximately 3 billion bases, the theoretical average coverage is:
Coverage = 30,000,000,000 / 3,000,000,000
= 10×
- This is an idealized estimate. Actual usable coverage may be lower because of sequencing errors, adapter contamination, low-quality reads, duplicate reads, contamination, and reads that cannot be confidently mapped to the reference genome.
- After sequencing, the instrument produces raw sequencing data. For short-read sequencing, these data are commonly represented in FASTQ files, which contain nucleotide sequences and corresponding quality scores. Long-read sequencing workflows may initially produce raw signal files or other platform-specific data before basecalling generates sequence reads. Oxford Nanopore sequencing can produce POD5 raw signal files, while PacBio workflows use their own data structures and commonly provide sequencing reads in BAM-based formats. These files may be converted or processed into formats suitable for downstream analysis.
- FASTQ files contain four lines for each sequence record: a sequence identifier, the nucleotide sequence, a separator line, and a string of quality scores. The quality scores represent the estimated confidence of individual base calls. They are commonly expressed using Phred quality scores, where higher values indicate lower estimated probabilities of base-calling errors. FASTQ data are important inputs for read quality assessment, filtering, alignment, and other sequencing analyses.
- The first major computational stage is sequencing quality control. It evaluates the quality of raw reads and identifies potential problems such as low-quality bases, adapter contamination, short reads, unusual nucleotide composition, and technical artifacts. FastQC is widely used for examining short-read sequencing quality. MultiQC combines reports from multiple tools into a consolidated summary. fastp can perform quality filtering, adapter trimming, and other preprocessing operations. For nanopore data, NanoPlot and NanoStat can summarize read length and quality distributions, while other platform-specific tools can be used for PacBio quality assessment.
- A simplified short-read quality-control workflow can be performed using the following commands:
# Examine raw FASTQ quality
fastqc sample_R1.fastq.gz sample_R2.fastq.gz
# Generate a combined quality-control report
multiqc .
# Filter reads and remove adapter sequences
fastp \
-i sample_R1.fastq.gz \
-I sample_R2.fastq.gz \
-o sample_R1_clean.fastq.gz \
-O sample_R2_clean.fastq.gz \
–html fastp_report.html \
–json fastp_report.json
- The filtering parameters should be selected according to the sequencing technology, library design, and analysis objective. Excessive trimming may remove useful sequence information, while insufficient filtering may allow technical artifacts to affect downstream analysis. Quality-control reports should therefore be reviewed rather than relying only on automated filtering.
- After quality control, sequencing reads can be analysed using reference-based alignment or de novo genome assembly. In reference-based analysis, reads are mapped to an existing reference genome. This approach is commonly used when a suitable reference genome is available and the primary objective is to identify genetic differences between the sequenced sample and the reference. In de novo assembly, reads are computationally connected to reconstruct the genome without requiring a pre-existing reference. This approach is particularly important for organisms with no suitable reference genome or when the goal is to develop an improved genome assembly.
- Reference-based alignment is performed using specialized sequence aligners. BWA-MEM2 is widely used for aligning short DNA reads to reference genomes. Bowtie2 is another short-read aligner, while minimap2 supports long-read alignment and several other sequence-mapping tasks. HISAT2 is commonly used for RNA-sequencing alignment rather than standard whole-genome DNA read alignment. The selected aligner should match the sequencing technology and data type.
- A simplified workflow for aligning paired-end short reads to a reference genome using BWA-MEM2 is:
# Align paired-end reads to a reference genome
bwa-mem2 mem reference.fa \
sample_R1_clean.fastq.gz \
sample_R2_clean.fastq.gz \
> sample.sam
# Convert SAM to BAM and sort the alignments
samtools view -b sample.sam | \
samtools sort -o sample.sorted.bam
# Index the sorted BAM file
samtools index sample.sorted.bam
# Generate alignment statistics
samtools flagstat sample.sorted.bam
- The reference genome should be indexed before alignment, using the indexing commands required by the selected aligner. Alignment results should be assessed for mapping rate, coverage distribution, read duplication, insert-size distribution, and other relevant metrics. A high mapping rate does not automatically guarantee a correct analysis, because contamination, reference bias, and misalignment in repetitive regions can still affect results.
- The alignment stage produces files such as SAM, BAM, or CRAM. SAM is a text-based format containing sequence alignment information, while BAM is its compressed binary equivalent. CRAM is a more compact format that uses a reference sequence to support efficient storage. These files contain information about read positions, mapping quality, alignment orientation, and other attributes. SAMtools and related utilities are commonly used to sort, index, filter, and inspect alignment files.
- Variant calling is one of the most important applications of whole-genome sequencing. It identifies genetic differences between the sequenced sample and a reference genome. These differences may include single-nucleotide variants (SNVs), insertions, deletions, copy-number variations, and larger structural variants. Small variants are often analysed using specialized variant callers, while structural variants may require additional tools designed to detect larger genomic changes.
- Single-nucleotide variants represent changes at individual nucleotide positions. For example, a reference nucleotide may contain adenine while the sample contains guanine at the corresponding position. Insertions and deletions, commonly called indels, involve the addition or removal of nucleotides. These variants can occur in coding regions, regulatory regions, introns, and intergenic regions. Their biological effects depend on genomic location, sequence context, and functional consequences.
- Common small-variant callers include GATK HaplotypeCaller, FreeBayes, and bcftools. GATK provides widely used workflows for germline variant discovery, while FreeBayes and bcftools provide alternative approaches for analysing sequencing data. DeepVariant is another variant-calling tool that uses machine-learning methods and supports suitable sequencing data types. Variant-calling methods should be selected according to the organism, sequencing technology, ploidy, sample type, and research objective.
- A simplified example of variant calling using bcftools is:
# Generate a pileup from aligned reads
bcftools mpileup \
-f reference.fa \
-Ou sample.sorted.bam | \
bcftools call \
-mv \
-Oz \
-o sample_variants.vcf.gz
# Index the compressed VCF file
bcftools index sample_variants.vcf.gz
- This is an illustrative workflow rather than a complete clinical or production-grade variant-calling pipeline. Real analyses require appropriate alignment filtering, sample-specific settings, variant quality assessment, normalization, and validation. Variant-calling parameters should be selected according to the experimental design and the documentation for the software version being used.
- Structural variant analysis is another important component of whole-genome sequencing. Structural variants include large deletions, insertions, duplications, inversions, translocations, and complex genomic rearrangements. Short-read sequencing can identify many structural variants, but its ability to resolve large or repetitive events may be limited. Long-read sequencing can improve structural variant detection by spanning breakpoints and providing extended sequence context. Tools such as Manta, DELLY, and LUMPY have been used for short-read structural variant analysis, while Sniffles2, cuteSV, and SVIM support suitable long-read datasets.
- Copy-number variation (CNV) analysis identifies gains or losses of genomic segments. CNVs may involve a single gene, multiple genes, or large chromosome regions. Whole-genome sequencing can detect CNVs by examining read depth, paired-end mapping patterns, and other alignment features. Tools such as CNVkit and Control-FREEC support copy-number analysis, while specialized methods may be required for different organisms and sequencing designs. CNV interpretation should consider genomic context, sample ploidy, sequencing coverage, and possible technical biases.
- Variant annotation is performed to determine the genomic location and potential functional consequences of identified variants. Annotation can establish whether a variant is located in an exon, intron, splice site, promoter, untranslated region, or intergenic region. It can also predict whether a coding variant produces a synonymous, missense, nonsense, or frameshift change. Tools such as Ensembl Variant Effect Predictor (VEP), SnpEff, and ANNOVAR are widely used for variant annotation. Annotation databases and transcript versions should be recorded because different releases may produce different results.
- For human genomic studies, variant interpretation may involve resources such as ClinVar, gnomAD, OMIM, ClinGen, and disease-specific databases. ClinVar provides submitted interpretations of variants in relation to health conditions, while gnomAD provides population variation data that can help evaluate allele frequencies. ClinGen supports gene–disease validity and clinical genomic interpretation resources. These databases provide useful evidence, but their entries may differ in quality, completeness, and interpretation. A database annotation alone does not establish that a variant causes a disease.
- Whole-genome sequencing can be used for both germline and somatic variant analysis. Germline variants are inherited or arise in reproductive cells and may be present in nearly all cells of an individual. They are important in inherited disease research, population genetics, ancestry studies, and evolutionary analysis. Somatic variants arise during an individual’s lifetime and may be restricted to particular tissues or cell populations. They are especially important in cancer genomics, where tumour sequencing can identify acquired mutations, structural changes, and genomic alterations associated with tumour development.
- Cancer whole-genome sequencing may involve sequencing both tumour tissue and matched normal tissue. Comparing the two datasets helps distinguish tumour-specific somatic variants from inherited germline variants. Tumour purity, subclonal populations, sequencing depth, and normal-cell contamination can influence variant detection. Analysis may include single-nucleotide variants, indels, structural variants, copy-number changes, mutational signatures, and complex rearrangements. The interpretation of cancer genomes often requires integration with clinical, pathological, and molecular evidence.
- Whole-genome sequencing is also important in population genomics and evolutionary research. By comparing complete genomes from multiple individuals or populations, researchers can investigate genetic diversity, population structure, demographic history, natural selection, migration, and evolutionary relationships. WGS provides information about both coding and noncoding variation, allowing researchers to examine a broad range of genomic differences. Population-level analysis requires careful sampling design and consideration of relatedness, ancestry, population stratification, and sequencing batch effects.
- In microbial genomics, whole-genome sequencing is used to characterize bacteria, archaea, viruses, fungi, and other microorganisms. It can support species identification, strain typing, outbreak investigation, antimicrobial resistance analysis, virulence profiling, and evolutionary studies. Whole-genome sequencing of bacterial isolates can provide higher-resolution comparisons than many traditional typing methods. However, genomic similarity alone does not establish direct transmission between individuals or locations; epidemiological data, sampling information, and timing must also be considered.
- In agriculture and animal breeding, WGS can be used to identify genetic variation associated with production traits, disease resistance, fertility, adaptation, and other economically important characteristics. Whole-genome data can support genome-wide association studies, genomic selection, breed characterization, parentage analysis, and the investigation of genetic disorders. In crop research, WGS contributes to the identification of genes associated with yield, quality, stress tolerance, and disease resistance. In livestock, it can help characterize genetic diversity and identify variants that may influence productivity, health, and reproductive performance.
- Whole-genome sequencing also supports functional genomics by providing information about genes, regulatory regions, and other genomic elements. However, DNA sequence data alone do not reveal every aspect of gene activity or regulation. RNA sequencing, chromatin accessibility assays, DNA methylation profiling, proteomics, and other omics approaches may be integrated with WGS to investigate how genomic variation affects biological processes. Such integrated analyses can connect genetic differences with molecular functions, gene expression, cellular responses, and phenotypic traits.
- De novo whole-genome assembly is particularly important when no suitable reference genome is available. In this approach, sequencing reads are assembled into contigs and scaffolds. Contigs are continuous assembled DNA sequences, while scaffolds connect contigs using additional information about their relative order and orientation. Short-read assemblers such as SPAdes are widely used for suitable microbial and other datasets. Long-read assemblers such as Flye, Canu, and hifiasm support different long-read data types. Chromosome-scale scaffolding may use Hi-C sequencing, optical mapping, or other long-range technologies.
- Genome assembly quality is assessed using several complementary measures. N50 describes the contig or scaffold length at which half of the total assembly length is contained in sequences of that length or longer. Although N50 is commonly reported, a high N50 does not necessarily mean that an assembly is accurate or complete. QUAST provides assembly statistics and comparison measures, while BUSCO evaluates the presence of expected conserved genes. Read mapping, contamination screening, structural validation, and comparison with independent datasets provide additional evidence about assembly quality.
- A further consideration in WGS analysis is the handling of repetitive DNA. Repetitive sequences include tandem repeats, transposable elements, satellite DNA, and segmental duplications. These regions can complicate read alignment and genome assembly because multiple genomic locations may contain similar sequences. Short reads may not provide sufficient information to identify the correct location of a repeated fragment. Long reads can improve resolution by spanning repeats and connecting them to unique flanking regions. Specialized repeat-analysis tools and genome-annotation methods may be needed to characterize these regions.
- Whole-genome sequencing generates large datasets and requires substantial computational infrastructure. Storage requirements depend on genome size, sequencing depth, read length, file format, and whether raw signal data and intermediate files are retained. Computational requirements include memory, processor time, and, for some basecalling and analysis methods, graphics processing unit resources. Efficient workflows use compressed files, indexed reference genomes, parallel processing, and workflow-management systems. Tools such as Snakemake and Nextflow can organize multi-step analysis pipelines and improve reproducibility.
- A simplified reference-based WGS workflow can be represented as follows:
Biological sample
|
v
DNA extraction and quality assessment
|
v
Library preparation
|
v
Whole-genome sequencing
|
v
Raw sequencing data
|
v
Read quality control and filtering
|
v
Reference genome alignment
|
v
Alignment processing and quality assessment
|
v
Variant calling
|
v
Variant filtering and annotation
|
v
- Biological interpretation and reporting
- For reproducible analysis, each stage should be documented. Important information includes the sample identifier, extraction method, library preparation protocol, sequencing platform, chemistry, read length, sequencing depth, and quality-control results. Software names, versions, parameters, reference genome assembly, annotation release, and variant-filtering criteria should also be recorded. The original sequencing files and intermediate analysis products should be retained when possible. Workflow scripts, configuration files, and checksums can help verify that data have not been unintentionally altered.
- Whole-genome sequencing has several advantages. It provides broad coverage of the genome, supports the detection of coding and noncoding variation, and enables the analysis of single-nucleotide variants, indels, structural variants, and copy-number changes. It can contribute to reference genome development, population studies, disease research, microbial surveillance, and agricultural genomics. The availability of both short-read and long-read technologies allows researchers to select or combine sequencing approaches according to their analytical requirements.
- Despite these advantages, whole-genome sequencing has limitations. It can be more expensive and computationally demanding than targeted sequencing, particularly when high coverage or long-read data are required. Large datasets require substantial storage and processing resources. Not all genomic regions can be sequenced, aligned, or assembled with equal reliability. Repetitive regions, highly homologous sequences, GC-rich regions, and complex structural variants may remain difficult to analyse. In addition, many variants have uncertain biological significance, particularly those located in noncoding regions. WGS therefore produces extensive genetic information, but interpretation requires appropriate databases, functional evidence, statistical methods, and biological knowledge.
- Human whole-genome sequencing also raises ethical, legal, and privacy considerations. A genome contains sensitive information about an individual and may reveal information relevant to biological relatives. Genetic data should therefore be collected, stored, shared, and analysed according to applicable consent requirements, privacy regulations, and research ethics standards. Clinical interpretation requires appropriate professional expertise, validated analytical procedures, and consideration of the limitations of genetic testing. Research findings should be distinguished from clinically established conclusions.
- Whole-genome sequencing is a foundational technology in modern bioinformatics because it enables comprehensive investigation of genetic information across an organism’s genome. By combining DNA extraction, library preparation, sequencing, computational alignment or assembly, variant detection, annotation, and biological interpretation, WGS provides a broad framework for understanding genetic variation and genome organization. Its applications range from inherited disease research and cancer genomics to microbial surveillance, evolutionary biology, and agricultural improvement. Understanding the sequencing technologies, coverage requirements, analytical tools, limitations, and reproducibility principles is essential for designing reliable whole-genome sequencing studies.