Next Generation Sequencing

Next generation sequencing (NGS) refers to high throughput DNA sequencing technologies that parallelize the sequencing process, producing millions to billions of reads simultaneously. This guide is for researchers, clinicians, and students who need a practical, source bounded framework for understanding and implementing NGS. The fundamental shift from Sanger sequencing to NGS is that NGS sequences many DNA fragments in parallel, drastically reducing cost and time per base while enabling applications such as whole genome sequencing, transcriptomics, and metagenomics. A solid grounding in core concepts, decision points, and common pitfalls helps you plan experiments and interpret results correctly. The authoritative technical references from the NCBI Bookshelf provide a foundation for the principles covered here.

To navigate NGS choices, you must consider throughput, read length, error profiles, and cost. For example, short read platforms such as those from Illumina dominate for variant detection and gene expression, while long read technologies like Oxford Nanopore and PacBio are preferred for structural variant calling and de novo assembly. The official training resources at EMBL-EBI Training offer structured courses that complement this guide and can deepen your understanding of specific platforms.

At a Glance

Decision Criteria for Choosing an NGS Approach

Selecting an NGS platform and experimental design depends on your biological question, sample type, and budget. A systematic evaluation of the following criteria helps avoid costly mistakes.

Read Length Requirements. If your goal is to detect single nucleotide variants (SNVs) in human genomes, short reads of 100 150 bp are sufficient. For resolving repetitive regions, phasing haplotypes, or assembling genomes de novo, long reads from PacBio or Oxford Nanopore are necessary. The open bioinformatics workflows from the Galaxy Training Network provide practical tutorials on aligning both short and long reads.

Throughput Needs. Consider how many samples you need to sequence and the depth required. Whole genome sequencing of a human genome at 30x coverage requires roughly 90 Gb of data. For targeted panels, throughput demands are lower. The Bioconductor project contains R packages for power analysis and sequencing depth estimation that are useful during planning.

Error Profile and Accuracy. Illumina platforms typically achieve error rates below 0.1%, making them ideal for clinical variant calling. Long read platforms have higher raw error rates (1 15%) but can be improved with circular consensus sequencing (PacBio HiFi) or basecalling algorithms. For applications requiring high accuracy, such as identifying somatic mutations, short reads remain the gold standard. A study on mitochondrial DNA mutations in hepatocellular carcinoma used NGS to identify prognostic markers, demonstrating the importance of accurate variant calling somatic pathogenic mitochondrial DNA mutations independently predict worse prognosis in hepatocellular carcinoma.

Cost per Sample. Library preparation, sequencing reagents, and bioinformatics analysis all contribute to total cost. Multiplexing (barcoding) multiple samples in a single run reduces per sample cost but may reduce coverage. Use cost calculators available from sequencing centers.

Turnaround Time. For urgent clinical applications, platforms with rapid run times such as Oxford Nanopore (real time data streaming) or Ion Torrent (few hours) may be preferred. For large population studies, Illumina runs typically take 1 3 days.

Practical Workflow or Implementation Sequence

A robust NGS workflow includes four main phases: experimental design, library preparation, sequencing, and data analysis. Each phase requires careful quality control.

Step 1: Experimental Design. Define your biological question and choose the appropriate assay: whole genome, exome, targeted panel, RNA seq, or epigenomic. Determine sample size, replication, and sequencing depth. For clinical studies, consult guidelines from reputable consortia. The NCBI Sequence Read Archive hosts metadata from thousands of studies that can help benchmark your design.

Step 2: Library Preparation. Extract DNA or RNA of high purity and integrity. Fragment the nucleic acids to the desired size (e.g., 200 600 bp for short read libraries). Ligate platform specific adapters that include sequencing primers, barcodes, and flow cell binding sequences. Amplify the library if required (e.g., for Illumina) or proceed with native DNA (for Nanopore). Validate library concentration and fragment size using a Bioanalyzer or a fluorometric method.

Step 3: Sequencing. Load the library onto the chosen platform. Monitor run metrics in real time: cluster density (Illumina), read quality scores, and yield. For clinical diagnostics, adhere to standardized protocols. An allele counting based approach for non invasive prenatal testing used NGS to accurately determine paternity in trisomy 21 cases, illustrating the need for meticulous sequencing conditions allele counting based non invasive prenatal paternity testing for trisomy 21 fetuses.

Step 4: Data Analysis. Demultiplex raw data to assign reads to samples. Perform quality control using tools like FastQC. Trim adapters and low quality bases. Align reads to a reference genome (e.g., using BWA for short reads or minimap2 for long reads). Call variants (SNVs, Indels, structural variants) and annotate them with databases like dbSNP or ClinVar. For RNA seq, quantify expression levels. Use reproducible workflows available on Galaxy or through Bioconductor packages.

Step 5: Interpretation and Validation. Filter variants based on quality, frequency, and predicted impact. Validate clinically relevant findings using orthogonal methods such as Sanger sequencing or digital PCR. Document all steps for reproducibility.

Quality Checks

Systematic quality checks at each stage prevent downstream errors and false discoveries.

Pre sequencing Quality. Assess DNA/RNA integrity using gel electrophoresis or a Bioanalyzer. Verify library concentration and size distribution. Use qPCR to quantify amplifiable library molecules. A common target is a sharp peak at the expected fragment length with no adapter dimers.

Sequencing Run Quality. Monitor Q score metrics. For Illumina, aim for >80% of bases above Q30. Check cluster density: too low reduces yield, too high causes signal overlap. Examine per base sequence content for bias. For long reads, evaluate read length N50 and throughput.

Post sequencing Quality. Use FastQC to check per base quality scores, GC content, duplication levels, and adapter contamination. Remove low quality reads and adapters. For alignment, check mapping rate (typically >90% for human genomes) and coverage uniformity. Use tools like Picard to calculate duplication rates. High duplication may indicate PCR bias or over sequencing.

Interpret quality metrics in the context of your experiment. For pathogen detection, lower coverage may still be sufficient. For somatic mutation calling, deep coverage of 500x or more may be required. A study on avian influenza used NGS to track viral evolution, emphasizing the importance of depth for detecting minority variants Rapid Expansion of Highly Pathogenic Avian Influenza A(H5N1) Clade 2.3.4.4b Genotype D1.1 Virus across Flyway Regions, North America, Fall 2024.

Common Mistakes

Avoid these frequent errors to improve the reliability of your NGS projects.

Insufficient Coverage Depth. Underpowered experiments fail to detect rare variants or low expression transcripts. Calculate required depth before starting. For human whole genome sequencing, 30x is standard for germline variants, 100x or more for somatic calling.

Ignoring Batch Effects. Process samples in a randomized order and use consistent reagents. Batch effects can introduce systematic bias that mimics biological signals. Include technical replicates and use statistical methods to correct for batch effects during analysis.

Poor Library Quality. Degraded DNA or RNA, adapter contamination, and uneven amplification all degrade data quality. Always validate libraries before sequencing. Do not overload the flow cell with adapter dimers.

Overlooking Bioinformatics Capabilities. NGS generates massive datasets that require substantial computational resources and expertise. Plan for storage, processing, and analysis before sequencing. Use cloud resources or local servers with sufficient RAM.

Misinterpreting Variant Frequencies. Population specific variants, sequencing artifacts, and alignment errors can lead to false positives. Use stringent filtering criteria and validate with orthogonal methods. For rare diseases, consider trio analysis to confirm inheritance patterns. The distinct genomic landscape of colorectal signet ring cell carcinoma revealed frequent KMT2 family alterations through careful NGS analysis and validation Distinct genomic landscape of colorectal signet ring cell carcinoma reveals frequent KMT2 family alterations and SMAD4 inactivation.

Limits of Interpretation

NGS data must be interpreted with caution due to several inherent limitations.

Technical Limitations. Short reads cannot uniquely align to repetitive regions or resolve complex structural variants. GC bias during PCR amplification can skew coverage. Long reads have higher error rates that may introduce false variants. Basecalling algorithms for Nanopore data continue to improve but may still produce systematic errors in homopolymers.

Biological Limitations. Mosaic variants present at low frequency require deep sequencing and careful calling. Somatic mutations in heterogeneous tissues may be missed. RNA seq data reflects steady state transcript levels and does not capture protein abundance or post translational modifications. For infectious disease diagnostics, NGS may not distinguish between viable and non viable pathogens. A granulomatous inflammation study used a molecular algorithm including NGS for etiologic diagnosis, highlighting the need to integrate sequencing with clinical and histopathological data Granulomatous inflammation in lung and lymph node specimens: A molecularly enhanced pathology based algorithm for etiologic differential diagnosis.

Interpretation Limitations. Variant classification requires population frequency data, functional predictions, and segregation information. Many variants are classified as variants of uncertain significance (VUS). Clinical actionability is often limited by incomplete knowledge. The application of artificial intelligence in clinical genetics is expanding but still faces challenges in validation and interpretability Artificial Intelligence in Clinical Genetics: Current Applications and Challenges.

Reproducibility. Different bioinformatics pipelines can yield different variant calls. Use standardized workflows, share code and parameters, and deposit raw data in public repositories like the NCBI Sequence Read Archive. Always report the version of software and reference genome used.

Frequently Asked Questions

What is the difference between NGS and Sanger sequencing? Sanger sequencing produces a single, high quality read per reaction, typically up to 1000 bp. NGS parallelizes millions of sequencing reactions, massively increasing throughput and lowering cost per base. NGS is suited for genome scale studies, while Sanger is still used for validating individual variants and sequencing small regions.

Can NGS detect structural variants? Yes, but sensitivity varies by platform. Short read NGS can detect deletions and duplications through read depth and paired end mapping, but it struggles with inversions and translocations. Long read NGS from PacBio or Oxford Nanopore excels at detecting large structural variants because reads span breakpoints. Combining both short and long reads provides the most comprehensive view.

How much sequencing data do I need? The required data amount depends on genome size, target depth, and application. For a human genome at 30x coverage, approximately 90 Gb of raw data is needed. For targeted panels, 1 5 million reads per sample may suffice. For RNA seq, 20 50 million reads per sample is typical for gene expression quantification. Use power calculations based on expected effect sizes.

What are the ethical considerations in NGS? NGS can reveal incidental findings, such as predisposition to diseases, non paternity, and carrier status. Informed consent should cover data sharing, return of results, and privacy protections. Researchers must comply with institutional review board requirements and data protection regulations. Depositing data in public repositories requires careful de identification.

References and Further Reading

• NCBI Bookshelf for authoritative technical references on sequencing technologies and molecular biology.
• EMBL-EBI Training for structured courses in bioinformatics and sequence analysis.
• Galaxy Training Network for open, hands on workflow tutorials.
• Bioconductor for software and documentation for genomic data analysis in R.
• NCBI Sequence Read Archive for public repository and metadata of high throughput sequencing data.
• Somatic pathogenic mitochondrial DNA mutations independently predict worse prognosis in hepatocellular carcinoma (Sci Rep) for an example of NGS in cancer genomics.
• Rapid Expansion of Highly Pathogenic Avian Influenza A(H5N1) Clade 2.3.4.4b Genotype D1.1 Virus across Flyway Regions, North America, Fall 2024 (Emerg Infect Dis) for NGS in pathogen surveillance.
• Allele counting based non invasive prenatal paternity testing for trisomy 21 fetuses (Leg Med (Tokyo)) for a clinical NGS application.
• Granulomatous inflammation in lung and lymph node specimens: A molecularly enhanced pathology based algorithm for etiologic differential diagnosis (Ann Diagn Pathol) for integrating NGS with histopathology.
• Distinct genomic landscape of colorectal signet ring cell carcinoma reveals frequent KMT2 family alterations and SMAD4 inactivation (BMC Cancer) for NGS in tumor profiling.
• Artificial Intelligence in Clinical Genetics: Current Applications and Challenges (Indian J Pediatr) for emerging tools in NGS interpretation.

Related Articles

• Protein Synthesis
• Incomplete Dominance Gene
• Cell Membrane Function Biology
• Dna Structure
• Protein Structure