- DNA sequencing is the process of determining the precise order of nucleotide bases—adenine (A), thymine (T), cytosine (C), and guanine (G)—in a DNA molecule. It is one of the most important technologies in modern molecular biology, genetics, genomics, and bioinformatics. DNA sequencing provides the information needed to study genes, identify genetic variants, characterize genomes, investigate evolutionary relationships, diagnose genetic diseases, understand cancer, monitor infectious organisms, and improve crops and livestock. Advances in sequencing technologies have made it possible to analyse individual DNA fragments, complete genomes, large populations, and complex biological communities at increasingly large scales.
- The development of DNA sequencing technologies has progressed from relatively slow methods that analyse individual DNA fragments to high-throughput platforms capable of generating millions or billions of sequence reads. Modern sequencing technologies are commonly grouped into three broad generations. First-generation sequencing includes classical methods such as Sanger sequencing. Second-generation sequencing, also called next-generation sequencing (NGS), includes high-throughput methods that generate large numbers of short DNA reads in parallel. Third-generation sequencing includes long-read technologies such as PacBio single-molecule real-time sequencing and Oxford Nanopore sequencing, which can analyse much longer DNA molecules and, in some applications, detect certain DNA modifications directly.
- The choice of sequencing technology depends on the biological question, genome size, required accuracy, read length, sequencing depth, sample quality, cost, throughput, and computational resources. Different technologies have different strengths and limitations. Short-read sequencing is highly effective for many applications involving accurate base identification and large-scale variant analysis, while long-read sequencing is particularly useful for genome assembly, structural variant detection, repetitive regions, haplotype resolution, and the characterization of complex genomic regions. Combining different technologies can provide a more complete understanding of genome structure and variation.
- The first major DNA sequencing method was developed by Frederick Sanger and colleagues and became widely known as the chain-termination method. Sanger sequencing uses DNA polymerase to synthesize a complementary DNA strand from a template. The reaction contains normal deoxynucleotide triphosphates (dNTPs) and fluorescently labelled or otherwise distinguishable dideoxynucleotide triphosphates (ddNTPs). Unlike normal nucleotides, ddNTPs lack the 3′-hydroxyl group required for the formation of the next phosphodiester bond. When a ddNTP is incorporated into a growing DNA strand, further extension stops.
- The resulting DNA fragments differ in length according to the position at which chain termination occurred. In automated Sanger sequencing, fluorescent labels distinguish the terminal nucleotide of each fragment. The fragments are separated according to size by capillary electrophoresis, and a detector records the fluorescent signals as they pass through the instrument. Software converts these signals into a chromatogram and a nucleotide sequence. The chromatogram contains coloured peaks representing the detected bases, and the quality of the peaks helps researchers assess the reliability of the sequence.
- Sanger sequencing generally produces reads of several hundred bases, commonly around 500–900 bases under suitable conditions, although read length and quality depend on the sample, chemistry, instrument, and analysis. It is highly useful for sequencing individual DNA fragments, validating selected variants, confirming cloned DNA constructs, and analysing targeted regions. It remains valuable in laboratories because it provides relatively straightforward sequence interpretation and is suitable when only a limited number of samples or regions need to be examined. However, it has relatively low throughput compared with modern NGS technologies and is not generally the most efficient approach for sequencing large genomes or thousands of samples.
- The second generation of DNA sequencing introduced high-throughput sequencing technologies that analyse many DNA molecules simultaneously. These methods are commonly called next-generation sequencing. Instead of sequencing one DNA fragment at a time, NGS platforms generate large collections of reads through parallel sequencing reactions. This approach greatly reduced the cost per base and made whole-genome sequencing, whole-exome sequencing, transcriptome analysis, metagenomics, and large population studies more practical.
- Most short-read NGS workflows begin with DNA extraction and quality assessment. The DNA is fragmented into smaller pieces, and sequencing adapters are added to the fragment ends. Adapters are short, known DNA sequences that allow fragments to attach to the sequencing platform and provide binding sites for amplification or sequencing primers. Sample-specific index sequences, also called barcodes, may be added so that multiple samples can be pooled and sequenced together. This process is known as library preparation. The quality and quantity of the library are assessed before sequencing.
- One of the most widely used second-generation technologies is sequencing by synthesis, particularly the Illumina approach. In this method, DNA fragments are attached to a flow cell and amplified to form clusters of identical or nearly identical DNA molecules. During sequencing, fluorescently labelled reversible-terminator nucleotides are incorporated one base at a time. After each incorporation, the instrument records the fluorescent signal to identify the added nucleotide. The fluorescent label and blocking group are then removed, allowing the next cycle of synthesis to proceed. Repeated imaging and base identification produce the DNA sequence.
- Illumina sequencing is widely used because it provides high-throughput sequencing with generally high per-base accuracy. It is applied in whole-genome sequencing, whole-exome sequencing, RNA sequencing, targeted gene panels, metagenomics, ChIP-seq, ATAC-seq, and many other genomic applications. Paired-end sequencing is commonly used, in which both ends of a DNA fragment are sequenced. This produces two reads associated with the same original fragment and provides additional information about the distance and orientation between them. Paired-end data can improve alignment, genome assembly, and the detection of certain structural variations compared with single-end data.
- Other second-generation technologies have also contributed to the development of high-throughput sequencing. The 454 pyrosequencing platform used the detection of light generated during nucleotide incorporation. When a nucleotide was incorporated by DNA polymerase, pyrophosphate was released and used in an enzymatic reaction that generated a light signal. The intensity of the signal was related to the number of identical nucleotides incorporated in succession. This technology helped establish early high-throughput sequencing, but difficulties in accurately determining the length of homopolymer regions and changes in platform availability limited its continued use.
- The SOLiD sequencing technology used sequencing by ligation rather than sequencing by synthesis. Fluorescently labelled oligonucleotide probes were ligated to a primer in a series of reactions, and the resulting signals were used to infer the nucleotide sequence. Its two-base encoding approach provided a particular strategy for error detection, but the workflow and data interpretation were more complex than those of some competing platforms. SOLiD played an important role in the history of NGS but is not a major choice for most current sequencing projects.
- Ion Torrent sequencing is another sequencing-by-synthesis approach. It detects hydrogen ions released when a nucleotide is incorporated into a growing DNA strand. These ions produce a change in pH that is detected electronically by the instrument. Unlike optical sequencing systems, this approach does not require fluorescent imaging. Ion Torrent technology has been used in targeted sequencing, clinical research, microbial genomics, and other applications. Its performance depends on the sequencing chemistry and instrument, and homopolymer regions can present challenges because signal intensity must be used to estimate how many identical bases were incorporated.
- Second-generation sequencing has transformed genomics by making it possible to sequence large numbers of DNA fragments in parallel. However, its relatively short reads can make some genomic regions difficult to analyse. Repetitive DNA sequences, segmental duplications, complex structural variants, and highly similar gene families can be challenging to assemble or map accurately using short reads alone. These limitations contributed to the development and adoption of third-generation sequencing technologies, which can sequence individual DNA molecules and generate much longer reads.
- Third-generation sequencing is generally characterized by single-molecule sequencing, in which DNA molecules are analysed without requiring the same type of clonal amplification used in many short-read workflows. Long-read technologies can generate reads spanning thousands or tens of thousands of bases, and some methods can produce reads extending beyond this range. Longer reads can bridge repetitive regions, connect distant genomic features, resolve complex structural variations, and improve the assembly of complete chromosomes. The exact read length, accuracy, and throughput depend on the platform, sequencing chemistry, sample preparation, and analysis method.
- PacBio single-molecule real-time sequencing (SMRT sequencing) is an important third-generation technology. It uses specialized structures called zero-mode waveguides (ZMWs), which allow the observation of DNA synthesis by an individual DNA polymerase molecule. A DNA polymerase is attached within a ZMW, and fluorescently labelled nucleotides are introduced into the reaction. When a nucleotide is incorporated, its fluorescent signal is detected. The label is associated with the phosphate group and is released during incorporation, allowing the polymerase to continue synthesizing the DNA strand.
- PacBio sequencing can generate long reads that are valuable for genome assembly, transcript characterization, structural variant analysis, and haplotype phasing. An important feature of PacBio technology is circular consensus sequencing (CCS), which produces highly accurate HiFi reads. In this approach, a DNA insert is incorporated into a circular template, allowing the polymerase to pass around the same molecule multiple times. The repeated observations are combined to generate a consensus sequence. HiFi reads provide a useful combination of long-read length and high accuracy, making them suitable for applications requiring reliable identification of single-nucleotide variants and complex genomic variation.
- Oxford Nanopore sequencing is another major third-generation technology. It determines DNA sequence by measuring changes in electrical current as a DNA molecule passes through a biological nanopore embedded in a membrane. A motor protein controls the movement of DNA through the pore, and groups of nucleotides within or near the pore influence the measured electrical signal. Computational models called basecallers interpret the signal and convert it into nucleotide sequences. Because the technology analyses individual DNA molecules, it can produce long reads and can be used for real-time sequencing.
- Oxford Nanopore platforms include portable devices as well as larger instruments designed for higher-throughput applications. Nanopore sequencing can be used for genome assembly, pathogen identification, metagenomics, transcript analysis, structural variant detection, and field-based sequencing. Its ability to generate very long reads can be particularly useful for resolving repetitive genomic regions and complex structural changes. Depending on the library preparation and analysis approach, nanopore sequencing can also provide information about certain DNA modifications, including methylation, from the original electrical signal. However, accuracy and performance depend on the chemistry, pore type, basecalling model, read quality, and analysis workflow.
- A major distinction between short-read and long-read sequencing is the relationship between read length and sequence accuracy. Short-read platforms commonly produce relatively short sequences with high per-base accuracy, while long-read platforms can span much larger genomic regions and improve the resolution of complex genomic structures. Modern long-read technologies have substantially improved their accuracy, and PacBio HiFi sequencing in particular combines long reads with high consensus accuracy. Nevertheless, researchers should evaluate the error profiles of the specific technology and chemistry rather than assuming that all long reads have the same accuracy characteristics.
- DNA sequencing technologies can also be compared according to whether they require amplification, the type of signal detected, and the biological information they preserve. Some platforms amplify DNA fragments before sequencing, while single-molecule platforms can analyse individual molecules directly. Optical systems detect fluorescence, semiconductor systems detect changes in electrical or chemical signals, and nanopore systems detect changes in ionic current. These differences influence library preparation, read characteristics, sequencing throughput, error patterns, and the types of biological questions that can be addressed.
- The choice of sequencing platform also depends on the amount of sequencing required. Sequencing depth, or coverage, describes how many times a nucleotide position is represented by sequencing reads. For example, 30× average coverage means that the total number of aligned bases corresponds to approximately thirty times the length of the reference genome. Higher coverage can improve confidence in variant detection and genome assembly, but the required depth depends on the application. Human germline whole-genome sequencing often uses substantially greater depth than some microbial genome projects, while rare variant detection, low-frequency somatic mutations, and metagenomic studies may require specialized depth and analysis strategies.
- The relationship between sequencing depth and read length is important. High coverage with short reads may provide reliable information about individual bases but may not resolve long repeats or complex rearrangements. Long reads can span these regions and provide structural information, although some applications benefit from additional sequencing depth to improve consensus accuracy and detect low-frequency variants. Combining short-read and long-read data can therefore provide complementary information. Short reads may support accurate base-level polishing, while long reads help establish the overall genomic structure.
- DNA sequencing begins with careful sample preparation. Biological material may come from blood, saliva, tissue, cultured cells, microorganisms, plants, environmental samples, or other sources. DNA extraction methods must be selected according to the sample type and the intended sequencing technology. The extracted DNA is assessed for concentration, purity, integrity, and contamination. Methods such as spectrophotometry, fluorometry, and electrophoresis can provide complementary information about DNA quality and quantity. High-molecular-weight DNA is particularly important for many long-read sequencing applications because fragmentation can reduce the length of the resulting reads.
- Library preparation converts extracted DNA into a form compatible with the sequencing platform. Depending on the technology, this may involve fragmentation, end repair, adapter ligation, barcode addition, size selection, amplification, or enrichment of target regions. Some projects use PCR amplification to increase the amount of DNA available for sequencing, while others aim to reduce amplification bias or preserve native DNA molecules. Targeted sequencing libraries may use hybridization capture or amplicon-based methods to enrich selected genomic regions. The choice of preparation method can influence coverage uniformity, sequence representation, error profiles, and the ability to detect certain types of variation.
- After sequencing, the instrument generates raw data that must be processed before biological interpretation. Short-read sequencing commonly produces FASTQ files containing nucleotide sequences and corresponding quality scores. Long-read platforms may generate FASTQ files, signal data, or platform-specific intermediate files. Basecalling converts raw instrument signals into nucleotide sequences and quality values. Demultiplexing separates reads according to their sample-specific barcode sequences, while adapter trimming and quality filtering remove or identify technical sequences and low-quality reads.
- Quality control is an essential part of sequencing data analysis. Tools such as FastQC can summarize short-read quality characteristics, including per-base quality, sequence composition, adapter content, and duplication levels. MultiQC can combine quality-control reports from multiple samples or analysis tools. Fastp supports quality filtering, adapter trimming, and other preprocessing tasks. For long-read data, tools such as NanoPlot and platform-specific quality-control utilities can summarize read length, quality, and sequencing yield. Quality-control results should be interpreted in the context of the library preparation method and biological sample rather than applying identical thresholds to every dataset.
- Sequence alignment is another major stage in bioinformatics analysis. Reads are mapped to a reference genome to identify their genomic locations. Short-read aligners such as BWA-MEM2 and Bowtie 2 are widely used for different types of sequencing data. Long-read aligners such as minimap2 support the mapping of long DNA and RNA reads. Alignment files are commonly stored in SAM or BAM format, while CRAM provides a compressed representation that can reduce storage requirements when the appropriate reference information is available. SAMtools and related utilities are used to sort, index, filter, and inspect alignment files.
- Genome assembly is an alternative approach in which sequencing reads are combined to reconstruct longer genomic sequences without relying entirely on a pre-existing reference genome. De novo assembly is especially useful for organisms that lack a suitable reference or when the aim is to characterize genome structure and variation that may be absent from the available reference. Short-read assemblers include SPAdes and related tools, while long-read and hybrid assembly approaches use tools such as Flye, Canu, hifiasm, and other specialized assemblers. The appropriate assembler depends on the read type, genome characteristics, sequencing depth, and desired assembly quality.
- Genome assembly quality is assessed using measures such as contig number, contig length, N50, total assembly length, and completeness. QUAST provides assembly statistics and comparisons, while BUSCO estimates completeness using conserved single-copy orthologs expected for a selected lineage. For microbial genomes, CheckM and CheckM2 can help evaluate completeness and contamination. No single statistic fully describes assembly quality. A genome may have a high N50 but still contain misassemblies, missing genes, contamination, or unresolved repetitive regions. Multiple quality measures should therefore be considered together.
- Variant calling is one of the major applications of DNA sequencing. It involves identifying differences between a sample and a reference genome or between multiple samples. Common variant classes include single-nucleotide variants (SNVs), small insertions and deletions (indels), copy-number variations, and larger structural variants. Short-read variant-calling workflows often use tools such as GATK, FreeBayes, bcftools, and related software. Structural variant analysis may require specialized tools such as Manta, DELLY, Sniffles, or cuteSV, depending on the sequencing data and analysis objectives. Variant calls are commonly stored in VCF or BCF format.
- DNA sequencing is also widely used in whole-genome sequencing (WGS), which aims to analyse nearly the entire genomic DNA content of an organism. WGS can identify coding and noncoding variants, structural changes, and other genomic features. It is used in human genetics, microbial genomics, population studies, cancer research, evolutionary biology, and agricultural genomics. The ability to analyse a broad range of genomic regions makes WGS powerful, although interpretation can be challenging because many detected variants have uncertain or unknown biological significance.
- Whole-exome sequencing (WES) focuses primarily on protein-coding regions of the genome, which are collectively called the exome. It typically uses hybridization-based capture or another enrichment strategy to select exonic DNA before sequencing. WES can be useful for identifying variants associated with rare inherited disorders and other conditions in which coding variants are important. However, it does not comprehensively capture all noncoding regulatory regions, and coverage may be uneven. Some exons and structurally complex regions may be difficult to analyse reliably.
- Targeted gene-panel sequencing examines a selected set of genes or genomic regions associated with a particular disease, biological pathway, or research question. It can provide high coverage of specific targets and may be more economical than sequencing an entire genome when the relevant genes are already known. Targeted panels are used in clinical diagnostics, cancer research, pharmacogenomics, and pathogen surveillance. Their main limitation is that variants outside the selected regions are generally not detected, and the panel design can influence which variant types are reliably identified.
- DNA sequencing also supports transcriptomic and epigenomic research. Although RNA sequencing primarily analyses RNA molecules, complementary DNA is often generated before sequencing on short-read platforms. Long-read RNA sequencing can help characterize complete transcript structures, alternative splicing, and transcript isoforms. DNA sequencing technologies can also be adapted to study DNA methylation and other epigenetic modifications. PacBio and nanopore methods can provide information about certain modifications under appropriate experimental and computational conditions, while bisulfite sequencing and other specialized approaches are used to study DNA methylation. The exact method must be selected according to the modification and biological question.
- Metagenomic sequencing analyses genetic material obtained directly from a sample containing multiple organisms. It can be used to characterize microbial communities in soil, water, food, human-associated environments, and other ecosystems. Shotgun metagenomic sequencing can provide information about microbial species, gene content, metabolic pathways, and antimicrobial-resistance determinants. Long-read sequencing can help reconstruct microbial genomes and link genes to particular organisms, while short-read sequencing provides high-throughput sequence data for many community-analysis workflows. Metagenomic results depend on sampling, DNA extraction, sequencing depth, reference database quality, and computational classification methods.
- DNA sequencing is central to population genomics and evolutionary studies. By sequencing many individuals, researchers can identify genetic variation within and between populations, investigate demographic history, estimate genetic diversity, detect signals of selection, and study relationships among populations. Whole-genome data can be used to construct phylogenetic trees, identify population structure, and examine gene flow. Tools such as PLINK, bcftools, ANGSD, IQ-TREE, and population-genetic analysis packages support different stages of these workflows. The interpretation of population-level patterns requires attention to sampling, relatedness, demographic history, and statistical assumptions.
- In agriculture, DNA sequencing supports the discovery of genes and variants associated with crop yield, disease resistance, stress tolerance, nutritional quality, growth, fertility, and other important traits. Whole-genome sequencing and targeted sequencing can identify useful genetic variation in crops and livestock. Sequencing data can also support genomic selection, marker discovery, genome-wide association studies, and the characterization of germplasm collections. Long-read sequencing is particularly useful for analysing complex plant genomes, structural variation, gene clusters, and regions that are difficult to resolve with short reads.
- Clinical genomics uses DNA sequencing to investigate inherited diseases, cancer, infectious diseases, and pharmacogenomic variation. Germline sequencing examines variants inherited through the reproductive line, whereas somatic sequencing investigates genetic changes acquired during an individual’s lifetime, including mutations that arise in cancer cells. Clinical interpretation requires careful evaluation of variant quality, population frequency, inheritance, functional evidence, phenotype, and established clinical guidelines. Sequencing results may identify pathogenic variants, likely pathogenic variants, variants of uncertain significance, or findings that do not explain the clinical condition. A variant of uncertain significance should not be treated as a confirmed disease-causing variant.
- Sequencing technologies also have limitations. Errors may arise during DNA extraction, library preparation, sequencing, basecalling, alignment, assembly, and variant calling. PCR amplification can introduce bias and duplicates, while short reads may map ambiguously to repetitive or duplicated regions. Long-read sequencing can resolve many difficult regions but may have technology-specific error patterns and different requirements for DNA quality and computational analysis. Contamination, sample mix-ups, low coverage, reference bias, and incomplete databases can also affect results. The presence of a detected sequence does not automatically establish its biological function or clinical significance.
- Ethical, legal, and data-protection considerations are particularly important when sequencing human DNA. Genomic information can reveal sensitive information about individuals and their biological relatives. Clinical and research projects must follow appropriate consent procedures, data governance policies, access controls, and relevant regulations. Researchers should also consider the representation of different populations in reference databases, because underrepresentation can reduce the accuracy of variant interpretation and the generalizability of genomic findings.
- Reproducibility is essential in sequencing research. Researchers should document the sample source, DNA extraction method, library preparation protocol, sequencing platform, chemistry, read configuration, sequencing depth, quality-control procedures, software versions, reference genome assembly, and analysis parameters. Raw reads, processed data, genome assemblies, and variant files should be stored in appropriate repositories when permitted. Examples of major sequence data repositories include the Sequence Read Archive (SRA), European Nucleotide Archive (ENA), and DNA Data Bank of Japan (DDBJ), which participate in the International Nucleotide Sequence Database Collaboration. Human genomic data may require controlled-access repositories because of privacy and consent restrictions.
- DNA sequencing technologies have transformed biological research by making it possible to examine genetic information at scales that were previously impractical. Sanger sequencing remains useful for targeted sequence analysis and validation, second-generation sequencing provides high-throughput and generally accurate short reads, and third-generation technologies offer long reads and new opportunities for analysing complex genomes and individual DNA molecules. The selection of an appropriate technology depends on the biological question, sample characteristics, required accuracy, read length, throughput, and available resources. Together, these technologies form the foundation of modern genomics, clinical genetics, evolutionary research, biotechnology, and bioinformatics.