- Short-read next-generation sequencing (NGS) is a high-throughput DNA sequencing approach that determines the nucleotide sequences of millions to billions of relatively short DNA fragments in parallel. Unlike Sanger sequencing, which generally analyses one DNA fragment per reaction, short-read NGS processes a large number of DNA molecules simultaneously. This parallel sequencing capability makes it possible to investigate entire genomes, exomes, transcriptomes, microbial communities, and genetic variation at a scale that would be impractical using traditional sequencing methods.
- Short-read sequencing commonly produces reads ranging from approximately 50 to 300 nucleotides, depending on the sequencing platform and configuration. Illumina sequencing-by-synthesis is the most widely used short-read technology. Other sequencing systems have also used short-read or high-throughput sequencing approaches, including semiconductor-based sequencing. The exact read length, output, accuracy, and chemistry depend on the instrument and sequencing technology. Short-read sequencing is particularly useful for applications requiring high accuracy, substantial sequencing depth, and efficient analysis of large numbers of samples.
- The fundamental principle of short-read NGS is massively parallel sequencing. A DNA sample is first converted into a sequencing library containing many DNA fragments. These fragments are attached to a solid surface or another sequencing support, amplified or otherwise prepared for signal detection, and sequenced simultaneously. During sequencing, the instrument detects signals associated with nucleotide incorporation or another nucleotide-identification process. Computational software converts these signals into nucleotide sequences, known as reads. The resulting reads are processed and analysed using bioinformatics tools.
- The workflow begins with the collection and extraction of DNA or RNA, depending on the biological material and research objective. For DNA sequencing, genomic DNA is isolated from cells, tissues, microorganisms, or other biological samples. For RNA sequencing, RNA is extracted and usually converted into complementary DNA (cDNA) before library preparation. The quality and quantity of the starting material are important because degraded or contaminated samples can affect library construction, sequencing performance, and downstream analysis.
- After extraction, the nucleic acid is assessed for concentration, purity, and integrity. Fluorometric methods, such as Qubit assays, are commonly used to measure DNA or RNA concentration, while spectrophotometric measurements can provide information about sample purity. Instruments such as the Agilent Bioanalyzer and TapeStation can be used to assess fragment-size distribution and nucleic acid integrity. For RNA sequencing, RNA integrity is particularly important because degradation can introduce biases into transcript abundance measurements and reduce the reliability of expression analysis.
- The next major step is library preparation. In short-read sequencing, a sequencing library is a collection of DNA fragments that have been modified to make them compatible with the sequencing platform. Genomic DNA may first be fragmented into smaller pieces using mechanical methods such as sonication or enzymatic methods. In some workflows, DNA fragmentation and adapter addition are performed together using transposase-based methods, commonly referred to as tagmentation.
- Adapters are short, designed DNA sequences attached to the ends of sample fragments. They contain sequences required for attachment to the sequencing platform, primer-binding sites, and other elements needed for sequencing. Many library preparation systems also introduce unique index sequences, often called barcodes, that allow multiple samples to be pooled and sequenced in the same run. After sequencing, these index sequences are used to identify the sample from which each read originated. Unique molecular identifiers (UMIs) may also be incorporated in some protocols to label individual original DNA or RNA molecules. UMIs can help distinguish independent molecules from PCR duplicates and support more accurate molecular counting.
- Library preparation may include end repair, A-tailing, adapter ligation, size selection, amplification, and cleanup. End repair prepares DNA fragment ends for adapter attachment, while A-tailing adds an adenine nucleotide to the 3′ ends in workflows that use adapters with complementary thymine overhangs. Adapter ligation joins the adapters to the DNA fragments. Size selection may be used to enrich fragments within a desired length range, and PCR amplification may increase the amount of library material available for sequencing. Some workflows reduce or eliminate amplification to limit PCR-associated biases and duplicate formation.
- The prepared library is quantified and assessed before sequencing. Library concentration and fragment-size distribution help determine whether the sample is suitable for loading onto the sequencing instrument. Incorrect library concentration can result in poor cluster formation or reduced sequencing efficiency. Libraries from multiple samples may then be pooled according to their concentration and desired sequencing depth. The pooled library is loaded onto the sequencing platform using the manufacturer’s recommended procedure.
- Illumina sequencing is based on sequencing-by-synthesis (SBS). In this approach, DNA fragments carrying platform-compatible adapters are attached to complementary oligonucleotides on a flow cell. The flow cell is a solid surface containing many attachment sites. Each DNA fragment binds to the surface through its adapter sequence. The attached fragments undergo bridge amplification, in which a DNA strand bends over and hybridizes to a nearby surface-bound oligonucleotide. DNA polymerase copies the fragment, producing a double-stranded structure. Repeated cycles of amplification generate clusters containing many copies of the same original DNA fragment.
- Cluster formation is important because the fluorescent signal from a single DNA molecule would generally be too weak for reliable detection in conventional Illumina sequencing systems. By generating many copies of the same fragment in a localized region, the instrument can detect the signal produced during nucleotide incorporation. Each cluster represents an amplified population derived from an original library fragment, although amplification and library preparation can introduce duplicates or other biases that must be considered during analysis.
- During sequencing-by-synthesis, DNA polymerase incorporates fluorescently labelled reversible-terminator nucleotides into the growing DNA strand. Each nucleotide type is associated with a detectable signal. The reversible terminator temporarily prevents the addition of another nucleotide during the same cycle. After incorporation, the instrument records the fluorescent signal to identify the added base. The fluorescent label and blocking group are then removed, allowing the next nucleotide to be incorporated. This process is repeated over successive sequencing cycles, generating the nucleotide sequence of each cluster.
- The sequence is determined by recording the order of nucleotide incorporation. Instrument software processes the optical signals and performs base calling, converting them into nucleotide letters such as A, T, G, and C. The result is a large collection of short DNA reads. The number of sequencing cycles and the chemistry used determine the read length. Depending on the platform and configuration, sequencing may produce single-end or paired-end reads.
- Single-end sequencing reads each DNA fragment from one end and produces one sequence read per fragment. Paired-end sequencing reads both ends of the same library fragment, producing two reads commonly designated R1 and R2. The two reads originate from opposite ends of the same DNA fragment, and their approximate orientation and separation provide additional information about the original fragment. Paired-end sequencing improves the ability to map reads across repetitive regions, detect insertions and deletions, identify structural variation, and assemble genomes. The actual distance between paired reads depends on the library insert size and sequencing configuration.
- For example, a paired-end sequencing run configured as 2 × 150 produces two reads of approximately 150 bases from each sequenced fragment. The total number of sequenced bases per fragment is therefore approximately 300, although the reads are separated by the unsequenced region between them. This distinction is important because paired-end sequencing does not necessarily produce one continuous read of 300 bases.
- Short-read sequencing platforms generate raw sequencing data that must be converted into standard bioinformatics file formats. FASTQ is the principal format used to store sequencing reads together with their quality scores. Each FASTQ record generally contains four lines: a sequence identifier, the nucleotide sequence, a separator line, and a string of quality characters. These quality characters encode Phred scores that indicate the estimated reliability of individual base calls.
- A simplified FASTQ record is shown below:
@READ_001
ATGCTAGCTAGCTACGATCG
+
IIIIIIIIIIIIIIIIIIII
- The sequence identifier identifies the read, the second line contains its nucleotide sequence, and the fourth line stores the encoded quality values. The example is illustrative; actual sequencing identifiers and quality strings depend on the instrument and data-processing pipeline.
- Phred quality scores are used to evaluate the probability of base-calling errors. The relationship is expressed as Q = −10 log10(P), where P is the estimated probability that a base call is incorrect. A Q20 score corresponds to an estimated error probability of 1%, while Q30 corresponds to an estimated error probability of 0.1%. Higher quality scores indicate lower estimated error probabilities. Quality scores are important because sequencing errors can affect read alignment, variant detection, genome assembly, and gene-expression estimates.
- The first bioinformatics step after sequencing is usually raw-read quality control. Tools such as FastQC, fastp, and MultiQC are commonly used to examine sequence quality, GC content, sequence length distribution, adapter contamination, duplicated sequences, overrepresented sequences, and other characteristics. FastQC generates quality-control reports for individual FASTQ files, while MultiQC combines reports from multiple samples into a consolidated summary. The fastp program can perform several preprocessing operations, including quality filtering, adapter trimming, and reporting.
- A typical FastQC command is:
fastqc sample_R1.fastq.gz sample_R2.fastq.gz \
–outdir fastqc_results
MultiQC can then combine the results:
multiqc fastqc_results \
–outdir multiqc_report
- If adapter contamination or low-quality sequence regions are detected, read preprocessing may be required. Adapter sequences can remain in reads when the original DNA fragment is shorter than the sequencing read length. Low-quality bases may also occur near read ends or throughout reads. Trimming and filtering remove or exclude unsuitable sequence regions, improving the reliability of downstream analysis. However, aggressive trimming can shorten reads, reduce coverage, and remove biologically meaningful sequence information. Preprocessing parameters should therefore be selected according to the experiment and data quality.
- The fastp tool can be used to perform quality filtering and adapter trimming on paired-end FASTQ files:
fastp \
-i sample_R1.fastq.gz \
-I sample_R2.fastq.gz \
-o sample_R1.trimmed.fastq.gz \
-O sample_R2.trimmed.fastq.gz \
–html fastp_report.html \
–json fastp_report.json
- After quality control and preprocessing, the reads are usually analysed according to the research objective. In reference-based genome analysis, short reads are aligned to a reference genome. Read alignment identifies the locations in the reference sequence to which individual reads correspond. This makes it possible to detect genetic variants, estimate sequence coverage, examine genomic regions, and investigate gene expression or other molecular features.
- Common short-read aligners include BWA-MEM2, Bowtie2, and HISAT2. BWA-MEM2 is widely used for mapping DNA sequencing reads to reference genomes. Bowtie2 supports efficient alignment of short DNA sequences and is used in several genomic and transcriptomic workflows. HISAT2 is a splice-aware aligner designed for RNA sequencing, allowing reads that span exon–exon junctions to be mapped to a reference genome.
- An example of aligning paired-end DNA reads using BWA-MEM2 is:
bwa-mem2 mem \
reference_genome.fa \
sample_R1.trimmed.fastq.gz \
sample_R2.trimmed.fastq.gz \
> sample.sam
- The resulting SAM file contains alignment information, including read identifiers, mapping positions, alignment flags, mapping quality, and other fields. SAM files can be large, so they are commonly converted into compressed BAM files or reference-based CRAM files. SAMtools is widely used for these operations.
samtools view -bS sample.sam > sample.bam
samtools sort sample.bam -o sample.sorted.bam
samtools index sample.sorted.bam
- The sorted BAM file can be used for downstream analyses, while the index enables efficient access to alignments in particular genomic regions. CRAM can provide more compact alignment storage by using a reference sequence during encoding and decoding. The reference genome must therefore be available and correctly identified when working with reference-dependent CRAM files.
- After alignment, researchers commonly examine mapping statistics, coverage, duplicate reads, and alignment quality. SAMtools, Picard, and other tools can calculate alignment metrics and identify potential problems. Mapping quality indicates the confidence of a read’s placement, although its precise interpretation depends on the aligner. Reads may fail to map because of sequencing errors, contamination, differences between the sample and reference genome, repetitive regions, or inadequate reference representation.
- One major application of short-read NGS is variant calling. Variant calling identifies differences between the sequenced sample and a reference genome. These differences may include single-nucleotide variants (SNVs), small insertions and deletions (indels), and, with suitable methods, some larger structural variants. Variant callers use aligned reads and evidence such as base quality, mapping quality, read depth, and allele balance to estimate whether a sequence difference is likely to represent a genuine variant.
- Common variant-calling tools include GATK, bcftools, and FreeBayes. GATK provides widely used workflows for germline and somatic variant analysis, while bcftools supports variant calling and manipulation of VCF and BCF files. FreeBayes performs haplotype-based variant detection. The appropriate caller and workflow depend on the organism, sequencing design, ploidy, variant type, and research objective.
- An illustrative bcftools workflow for variant calling from a sorted and indexed BAM file is:
bcftools mpileup \
-f reference_genome.fa \
sample.sorted.bam \
-Ou |
bcftools call \
-mv \
-Oz \
-o sample_variants.vcf.gz
bcftools index sample_variants.vcf.gz
- The resulting VCF file stores variant information, including genomic coordinates, reference and alternate alleles, quality information, and sample genotypes when available. Variant annotation tools such as SnpEff, Ensembl VEP, and ANNOVAR can help predict the possible effects of variants on genes, transcripts, and proteins. Annotation does not by itself establish whether a variant causes disease or affects a biological trait; interpretation requires additional evidence.
- Short-read NGS is also widely used in RNA sequencing (RNA-seq), which investigates gene expression and transcript composition. In a typical RNA-seq workflow, RNA is extracted, converted into cDNA, fragmented or otherwise prepared, and sequenced. Reads may be aligned to a reference genome using splice-aware aligners such as STAR or HISAT2, or quantified using transcriptome-based approaches. Tools such as featureCounts and HTSeq can count reads assigned to genes, while Salmon and kallisto estimate transcript abundance using lightweight or pseudoalignment-based approaches.
- RNA-seq analysis can identify genes with different expression levels between biological conditions. Differential expression tools such as DESeq2, edgeR, and limma-voom use statistical models to compare expression measurements while accounting for aspects of experimental design and biological variability. Proper replication, normalization, batch-effect assessment, and multiple-testing correction are important for reliable conclusions. Short-read RNA sequencing is also used to investigate alternative splicing, although the short length of reads can make it difficult to reconstruct complete transcript isoforms, especially in genes with complex exon structures.
- Another important application is whole-genome sequencing (WGS), in which DNA from an organism is sequenced to investigate genetic variation across the genome. Short-read WGS can support studies of inherited disorders, population genetics, evolutionary relationships, cancer genomics, and microbial genome analysis. Whole-exome sequencing (WES) focuses on protein-coding regions and is often used to investigate variants in genes associated with inherited diseases. Targeted sequencing examines selected genomic regions, such as disease-associated genes, cancer-related genes, or specific genomic loci. These approaches differ in their target regions, sequencing depth, cost, and sensitivity to different types of genetic variation.
- Short-read NGS is also important in metagenomics, where DNA from a microbial community is sequenced without necessarily isolating each organism. Shotgun metagenomic sequencing can identify microbial species, estimate their relative representation, and investigate functional genes within a community. Tools such as Kraken2, Bracken, MetaPhlAn, and HUMAnN support taxonomic and functional profiling. Genome assembly tools such as MEGAHIT and metaSPAdes can reconstruct longer sequences from metagenomic reads. However, closely related organisms, uneven abundance, repeated sequences, and incomplete reference databases can complicate the interpretation of metagenomic results.
- In microbial genomics, short-read sequencing is frequently used for bacterial genome sequencing, pathogen surveillance, antimicrobial resistance research, and outbreak investigation. Reads can be mapped to a reference genome or assembled de novo. Tools such as SPAdes can assemble short reads into contigs, while QUAST can assess assembly statistics. Prokka and Bakta can support microbial genome annotation, and tools such as Snippy can be used in suitable reference-based bacterial variant-analysis workflows. For phylogenetic analysis, researchers may compare core-genome variants or conserved gene sequences using appropriate alignment and tree-building methods.
- Short-read sequencing can also support genome assembly. In de novo assembly, reads are reconstructed into longer sequences without relying on a complete reference genome. Many short-read assemblers use de Bruijn graphs, in which sequences are represented through overlapping k-mers. The assembler identifies relationships among these k-mers and reconstructs longer genomic sequences. Tools such as SPAdes, MEGAHIT, and ABySS are used in different assembly contexts. The quality of an assembly depends on sequencing depth, read quality, genome complexity, repeat content, heterozygosity, and the characteristics of the sequencing library.
- Although short-read sequencing is highly effective for many applications, it has important limitations. Short reads may not span long repetitive regions, making it difficult to determine the correct arrangement of repeated sequences. This can result in fragmented genome assemblies, unresolved structural variation, and ambiguous mapping. Closely related paralogous genes and duplicated genomic regions may also be difficult to distinguish when reads map equally well to multiple locations.
- Short-read sequencing can be affected by GC-content bias, PCR amplification bias, uneven coverage, sequencing errors, and contamination. PCR duplicates can distort estimates of sequencing depth and allele frequency if they are not appropriately identified or handled. In RNA-seq, transcript abundance estimates may be affected by transcript length, sequence composition, library preparation, and mapping ambiguity. In metagenomics, differences in organism abundance and reference database completeness can influence taxonomic and functional estimates.
- Another limitation is that short-read sequencing generally provides less direct information about long-range genomic structure than long-read technologies. Although paired-end reads and specialized library strategies provide information about fragment orientation and approximate insert size, they may not fully resolve large insertions, complex rearrangements, repetitive regions, or complete haplotypes. Long-read technologies such as PacBio HiFi and Oxford Nanopore sequencing can provide longer individual reads and may be more suitable for resolving some of these challenges. Hybrid approaches combine short-read accuracy and depth with long-read structural information.
- The choice of sequencing depth depends on the organism, genome size, sample type, experimental design, and analytical objective. Coverage is commonly expressed as the average number of times a reference position is represented by sequencing reads. For example, 30× coverage indicates that the total number of aligned bases is approximately 30 times the size of the reference genome, averaged across the genome. This does not mean that every position has exactly 30 reads, because coverage is usually uneven. Higher sequencing depth may improve the detection of low-frequency variants or provide more complete representation, but additional sequencing does not automatically overcome mapping ambiguity, systematic errors, or inadequate sample preparation.
- Reproducibility is an essential part of short-read NGS analysis. Researchers should record sample preparation methods, library preparation protocols, sequencing platform and configuration, read length, indexing strategy, and sequencing depth. They should retain raw FASTQ files, quality-control reports, reference genome accession and version, software versions, command-line parameters, and intermediate analysis files where practical. Reference genome versions are especially important because genomic coordinates can differ between assemblies. Variant records should be interpreted using clearly identified reference sequences and appropriate genome builds.
- Short-read next-generation sequencing has transformed modern bioinformatics by making large-scale DNA and RNA analysis faster and more accessible. Its combination of massively parallel sequencing, high-quality short reads, and mature computational workflows supports research in genomics, transcriptomics, metagenomics, molecular diagnostics, agriculture, evolutionary biology, and precision medicine. Understanding library preparation, sequencing-by-synthesis, paired-end reads, quality control, alignment, variant calling, and downstream interpretation provides a foundation for working with modern high-throughput sequencing data. Although long-read technologies have expanded the possibilities of genome analysis, short-read NGS remains a central technology because of its accuracy, throughput, established analytical methods, and broad range of applications.