Full Genome Sequencing
Full genome sequencing is the laboratory process that determines the complete DNA sequence of an organism's genome at a single time. It provides a base by base readout of nearly every nucleotide, enabling researchers to identify variants, study gene function, and track evolutionary relationships. This guide is for molecular biologists, bioinformaticians, genetic counselors, and graduate students who need a practical, source bounded framework for planning, executing, and interpreting full genome sequencing projects. Whether you are new to the technique or looking to refine your workflow, the following sections cover core concepts, decision points, implementation steps, quality checks, common mistakes, and limits of interpretation. For a thorough technical introduction, consult the free textbooks available through the NCBI Bookshelf. Additional training materials from the EMBL-EBI Training portal can help you navigate the bioinformatics side of analysis.
At a Glance
| Aspect | Key Information |
|---|---|
| Definition | Determination of nearly all DNA base pairs in a genome, including coding and noncoding regions. |
| Typical Output | 30x to 60x coverage for human genomes, with millions to billions of reads. |
| Common Platforms | Illumina (short reads, high accuracy), Pacific Biosciences (long reads, high error rate), Oxford Nanopore (long reads, real time). |
| Primary Applications | Variant discovery, population genomics, pathogen surveillance, cancer genomics, de novo assembly. |
| Main Challenges | Sample quality, coverage depth, read alignment errors, variant filtering, and assembly of repetitive regions. |
Core Concepts and Decision Points
Before starting a full genome sequencing project, you need to understand a few core concepts and make deliberate choices that will shape your results.
Genome coverage depth refers to the average number of times each nucleotide is sequenced. A depth of 30x is often considered the minimum for reliable variant calling in human genomes, but lower depths may suffice for some microbial genomes. The trade off between cost and confidence is governed by the scale of your study. A population genetics study on marine bivalves, for example, requires careful balancing of coverage and filtering stringency, as described in a recent work on variant filtering in scallops [7]. You should also consider whether you need a reference genome. If one exists, you can align reads to it (resequencing). If not, you must perform de novo assembly, which demands longer reads and higher coverage.
Read length is a critical decision point. Short reads (100-300 base pairs from Illumina) are highly accurate and cheap but struggle with repetitive regions and structural variants. Long reads (10,000+ base pairs from PacBio or Oxford Nanopore) resolve repeats and large structural changes but have higher error rates. Many modern projects combine both strategies. For clinical applications, accuracy is paramount. A recent study on JAG1 functional data showed that likelihood based calibration improves variant classification, highlighting the need for high quality sequencing data [8].
Platform selection depends on your budget, turnaround time, and biological question. Illumina dominates resequencing projects because of its low cost per base and high throughput. Nanopore offers portability and real time analysis, which is valuable for field epidemiology. For a recent example of pathogen genomics, a study on recombinant PEDV strains used full genome sequencing to characterize virulence [6]. The choice of platform affects not only the raw data but also the downstream bioinformatics tools.
The Galaxy Training Network provides hands on tutorials for comparing sequencing platforms and setting up analysis pipelines. Use these resources to test your decision criteria before committing to a large scale project.
Practical Workflow or Implementation Sequence
A typical full genome sequencing workflow follows these eight steps. Each step requires quality checks to avoid costly mistakes.
Sample preparation and DNA extraction. Obtain high molecular weight DNA. Use a method that minimizes fragmentation and contamination. Quantify with fluorometry and check purity with spectrophotometry. A 260/280 ratio near 1.8 and 260/230 ratio near 2.0 is desirable.
Library construction. Fragmentation (for Illumina) or shearing (for long reads) must be reproducible. For short read libraries, ligate adapters and amplify carefully to avoid PCR duplicates. For long read libraries, follow the manufacturer's specific protocol. Always include a positive control (e.g., a well characterized genome) and a negative control (no DNA).
Sequencing. Run your chosen platform according to manufacturer recommendations. Monitor run metrics: cluster density for Illumina, pore occupancy for Nanopore, and polymerase read length for PacBio. Stop and reassess if quality metrics fall below thresholds.
Base calling and quality scoring. Modern instruments perform base calling automatically. However, you can improve accuracy using tools like Guppy for Nanopore or the native Illumina software. The output is a set of reads in FASTQ format, each with a quality score per base.
Read quality assessment and preprocessing. Use FastQC or similar to check for adapter contamination, GC bias, and per base quality. Trim low quality bases and remove adapters with tools like Trimmomatic or Cutadapt. Reassess after trimming. This step is crucial, poor preprocessing leads to erroneous variant calls. The Galaxy Training Network offers detailed workflows for quality control.
Read mapping or de novo assembly. If a reference genome is available, map reads using a short read aligner (e.g., BWA) or long read aligner (e.g., Minimap2). If not, assemble reads de novo with tools like SPAdes (short reads) or Flye (long reads). Evaluate assembly statistics: N50, genome completeness, and number of contigs.
Variant calling and filtering. For resequencing, use a variant caller such as GATK HaplotypeCaller or FreeBayes. Apply hard filters based on quality, depth, and strand bias. For population genomic analyses, be mindful of the filtering trade offs that affect downstream demographic inferences, as discussed in the scallop study [7]. For clinical variant interpretation, functional data and likelihood based calibration can refine classification [8].
Annotation and interpretation. Annotate variants with tools like SnpEff or ANNOVAR. Compare against public databases (dbSNP, ClinVar). For structural variants, use dedicated callers such as Manta or Sniffles. The Bioconductor project offers R packages for downstream statistical analysis and visualization.
Quality checks should be integrated at every step. Verify mapping statistics (percentage of reads mapped, coverage uniformity). Check that your variant calls have a reasonable transition/transversion ratio (approximately 2.0 for human genomes). If something looks off, go back and troubleshoot the library preparation or bioinformatics pipeline.
Common Mistakes
Several recurring errors can compromise a full genome sequencing project.
Insufficient coverage depth. Researchers sometimes cut coverage to save money, only to miss heterozygous variants or fail to assemble repetitive regions. Always run a coverage simulation or pilot experiment. Use the NCBI Sequence Read Archive (SRA) to find studies with similar organisms and note the coverage used.
Ignoring GC bias. Library preparation and sequencing can systematically underrepresent GC rich or GC poor regions. Use normalization methods or long reads to mitigate this. If you do not correct for bias, you may miss important regulatory elements.
Neglecting sample quality. Degraded DNA produces short fragments that limit read length and mapping accuracy. For long read platforms, high molecular weight DNA is essential. A simple gel check or Bioanalyzer trace can save weeks of troubleshooting.
Overfiltering or underfiltering variants. In population genetics, overly aggressive filtering can remove rare but real variants, while lax filtering inflates false positives. The scallop study [7] illustrates how different filtering strategies alter inference. Base your filters on empirical data and, if possible, validate a subset of calls with Sanger sequencing.
Using the wrong reference genome. An incorrectly chosen or outdated reference leads to mapping artifacts. Always use the latest assembly and rename chromosomes to match your organism. For non model organisms, consider building a custom reference or using a closely related species.
Skipping reproducibility practices. Not documenting software versions, parameters, and data provenance makes it impossible to reproduce or publish your analysis. Use version control and containerized environments (e.g., Docker, Singularity). The EMBL-EBI Training includes modules on reproducible bioinformatics.
Limits and Interpretation Uncertainty
Full genome sequencing is powerful, but it does not produce a perfect genome. Understanding the limits helps you avoid overinterpreting results.
Incomplete assembly of repetitive regions. Centromeres, telomeres, and ribosomal DNA clusters are notoriously difficult to sequence with short reads. Even long read technologies struggle with extremely long repeats. The human genome reference has only recently been completed for these regions. For other species, many gaps remain.
Structural variant detection limitations. While long reads improve structural variant detection, small insertions or deletions in repetitive contexts remain challenging. The sensitivity of your pipeline depends on read length, coverage, and the variant type. A study on HIV integration sites used full genome sequencing to quantify proviral DNA, but relied on specialized methods for localization [9]. This underscores that not all genomic features are equally accessible.
Variant annotation uncertainty. Many variants fall in noncoding regions with unknown function. Computational predictions are not experimental validations. A variant classified as "likely pathogenic" may later be reclassified. The study on JAG1 functional data [8] shows how experimental calibration can improve classification, but such data is unavailable for most genes.
Methylation and epigenetics not captured. Standard whole genome sequencing does not preserve epigenetic marks. For methylation analysis, you need bisulfite sequencing or enzymatic conversion methods. Do not assume that a base change is functionally relevant without epigenetic context.
Biases in public databases. Variant databases are skewed toward populations of European descent, which can lead to incorrect interpretation of variants in underrepresented groups. Carefully consider your study population when using these resources.
Bioinformatics pipeline variability. Different tools and parameter sets can yield different variant lists. Always assess concordance across multiple callers and report your pipeline in detail. The Bioconductor community provides standardized workflows that reduce some of this variability.
For a recent application of full genome sequencing in virus surveillance, a study tracking influenza re emergence in Australia used genomic epidemiology to reconstruct transmission chains [11]. The authors highlight that the resolution of such inference depends on sampling density and genomic coverage. Incomplete sampling can miss important lineages.
Frequently Asked Questions
What is the difference between whole genome sequencing and whole exome sequencing? Whole genome sequencing covers nearly all DNA, including introns and intergenic regions, while whole exome sequencing only targets the protein coding exons (about 1-2% of the genome). Exome sequencing is cheaper but misses regulatory and noncoding variants that may be relevant.
How much data do I need for a full genome sequencing project? For a human genome at 30x coverage, you need roughly 90 billion base pairs of data. This translates to about 300 million paired end reads for a 150 bp read length. For bacteria, 100x coverage might require only tens of millions of base pairs. Use coverage calculators available at the NCBI SRA to estimate.
Can full genome sequencing detect structural variants like inversions or translocations? Yes, but detection depends on read length and the bioinformatics tool. Short reads often miss large structural variants unless they are flanked by unique sequences. Long reads and linked read technologies significantly improve detection. The Galaxy Training Network has tutorials on structural variant calling.
What are the main ethical considerations for human whole genome sequencing? Privacy, consent, and return of results are major concerns. Sequencing produces incidental findings, such as pathogenic variants in disease genes. Institutional review boards and research protocols must address data sharing, deidentification, and whether to disclose actionable findings. Consult your institution's ethics board before starting.
References and Further Reading
NCBI Bookshelf , Free, peer reviewed textbooks on molecular biology and genetics. Covers fundamentals of DNA sequencing. NCBI Bookshelf
EMBL-EBI Training , Official training resources for sequence analysis, including whole genome sequencing tutorials and data management. EMBL-EBI Training
Galaxy Training Network , Open access hands on tutorials for bioinformatics workflows, from quality control to variant calling. Galaxy Training Network
Bioconductor , Open source software packages for genomic analysis in R, including workflows for variant annotation and visualization. Bioconductor
NCBI Sequence Read Archive (SRA) , Public repository for raw sequencing data, useful for benchmarking and finding relevant datasets. NCBI SRA
Isolation and pathogenicity of a highly virulent recombinant GIIc subtype PEDV strain , Demonstrates full genome sequencing in viral pathogen characterization. PubMed 42443885
Navigating tradeoffs in variant filtering for population genetic and demographic inferences in non model marine bivalves , Practical guide on filtering decisions. PubMed 42443760
Likelihood based calibration improves the clinical utility of JAG1 functional data for variant classification , Shows how experimental data improves variant interpretation. PubMed 42442366
Quantification and localization of integrated HIV 1 in memory and naive CD4+ T cells , Example of full genome sequencing for viral integration analysis. PubMed 42441657
caRBP Pred: Leveraging Protein Language Models for Prediction of Chromatin Associated RNA Binding Proteins , Illustrates computational approaches complementary to genomic sequencing. PubMed 42441082
Respiratory virus genomic epidemiology during post pandemic re emergence of influenza in Australia , Case study using whole genome sequencing for public health surveillance. PubMed 42438688