We select sequencing coverage for WGS according to the study objective: 0.5-5× may be suitable for population studies with appropriate statistical analysis, 15× sometimes offers a compromise between cost and accuracy, and 30× is a common starting point for variant identification in a single individual. Sequencing coverage of approximately 100× is considered, for example, in cancer studies when the variants of interest are present in only some DNA molecules. The final choice also depends on the organism, ploidy, material quality and expected analytical sensitivity.
The comparison below primarily covers short-read WGS with mapping to the reference genome, e.g. Illumina PE150. Projects involving the assembly of genomes de novo, long-read sequencing and polyploid analysis require separate parameter selection.
What does genome coverage mean?
30× coverage means that each position in the genome is read an average of 30 times. However, this does not mean that every fragment will be represented by exactly 30 reads (some regions will have higher sequencing coverage, while others will have lower sequencing coverage or remain unsequenced).
When assessing results, it is therefore also worth considering coverage uniformity and the proportion of the genome reaching a specified coverage threshold, e.g. at least 10× or 20×.
Comparison: which sequencing coverage suits which project?
| Sequencing coverage | When is it worth considering? | Main benefit | Key limitation |
|---|---|---|---|
| 0,5× | Large population studies; profiling large-scale copy number changes in an appropriately designed study | Low sequencing cost per sample | Many positions lack direct read coverage; insufficient on its own for accurate genome-wide genotyping |
| 5× | Analysis of population diversity and structure; genotyping supported by models that account for genotype uncertainty | Ability to study more individuals | Limited confidence in genotype calls, particularly for heterozygous genotypes |
| 15× | Resequencing projects where lower sensitivity than at 30× is acceptable | Compromise between data output and cost | Greater risk of missing variants in regions with lower sequencing coverage |
| 30× | Identification of SNPs and small InDel variants in individual diploid organisms | Greater confidence in genotype calls and less reliance on imputation | Does not guarantee the detection of all variants or complete sequencing of repetitive regions |
| 100× | Selected cancer studies and analyses of variants with a reduced variant allele fraction | More reads that can support a variant | Higher cost; sensitivity remains limited at very low variant allele fractions |
This is an indicative guide to choosing sequencing coverage, not a set of universal quality thresholds. At low sequencing coverage, the analysis method is key; at higher levels, mapping quality and the nature of the changes of interest are key.
When should you choose lower or higher sequencing coverage?
0.5× or 5×: when a large number of samples is the priority
0.5× is worth considering, when you are planning a large cohort and have access to an appropriate reference panel for statistical imputation of missing or uncertain genotypes. The effectiveness of this approach depends on how well the panel matches the study population and on the frequencies of the variants being analysed.
5× is worth considering, when you need more direct sequence observations but the size of the study population remains the priority. The analysis should account for genotype uncertainty (simply increasing sequencing coverage from 0.5× to 5× does not remove all the limitations of low-pass WGS).
The absence of a reference panel does not always rule out statistical imputation of missing or uncertain genotypes, but working without one requires a sufficiently large sample set, appropriate methods and validation. Do not automatically extrapolate results obtained in humans to non-model species.
15× or 30×: when an individual's genotype matters
15× is an option, if your project allows for some missing or less certain genotype calls and the savings will allow you to increase the number of biological samples.
30× is the most commonly chosen starting point, when the priority is to distinguish confidently between homozygotes and heterozygotes and to identify variants without relying heavily on statistical imputation of missing or uncertain genotypes. For human WGS, typical recommendations are approximately 30-50×, depending on the application and analysis model.
30× or 100×: when a variant is present in only a proportion of the DNA
In a tumour sample, the variant allele fraction depends on factors including the tumour cell fraction, subclonality and copy number changes. Such projects therefore often require higher sequencing coverage and a matched normal sample.
100× does not guarantee detection of variants with a variant allele fraction of 1%. At a local sequencing coverage of 100×, such a variant yields an average of just one read supporting the alternate allele. This is an illustrative calculation, not a detection threshold; at very low variant allele fractions, deep targeted sequencing is worth considering.
Sample requirements
The requirements for standard Illumina WGS include:
| Parameter | Requirement |
|---|---|
| Material | Genomic DNA - gDNA |
| Total amount of DNA | >50 ng |
| Concentration | >10 ng/µl |
| Volume | >20 µl |
| Purity (A260/A280) | 1,7- 2,2 |
The requirements for amount, concentration and volume must all be met. We recommend sending additional material for quality control and library preparation. Details can be found in the description of the WGS service.
Lower sequencing coverage does not automatically mean that less DNA is required: input material requirements depend primarily on the library preparation method. For degraded DNA, FFPE, cfDNA or small amounts of material, the requirements must be determined on a case-by-case basis. Available protocols differ in their input requirements.
How much data do you need?
The basic calculation is simple:
Amount of data [Gb] = reference genome size [Gb] × target coverage
For a diploid organism, sequencing coverage is normally based on the length of the haploid genome; we do not multiply this length by two again.
Below are example calculations for a genome size of 3.0 Gb and a read configuration of PE150, i.e. 2 × 150 bp:
| Sequencing coverage | Theoretical data output | Number of PE150 read pairs |
|---|---|---|
| 0,5× | 1.5 Gb | 5 million |
| 5× | 15 Gb | 50 million |
| 15× | 45 Gb | 150 million |
| 30× | 90 Gb | 300 million |
| 100× | 300 Gb | 1 billion |
Our calculations; values before accounting for data losses.
One PE150 read pair nominally provides 300 bases. Gb means one billion bases, while GB means a gigabyte of memory; these units are not interchangeable.
In practice, you need to account for reads discarded during quality control, unmapped reads and duplicate reads. For example, assuming 80% of the sequencing data are usable, obtaining 90 Gb of usable sequence requires approximately 112.5 Gb of raw sequencing data. This is an example calculation; the actual amount of additional data needed depends on the sample and library.
Before placing an order, clarify whether the quotation specifies the amount of raw sequencing data, the amount of cleaned sequencing data or the target coverage after mapping and filtering.
Common mistakes when choosing sequencing coverage
What information should you include in your brief?
To select an appropriate method and data output, provide:
Describe your project and we will help you choose the method and data output
Send us information about the organism, samples and study objective. We will help you choose the technology, sequencing coverage and scope of bioinformatics analysis to suit the question your project aims to answer.
Selected projects and client publications
See which institutions we perform next-generation sequencing (NGS) for and explore scientific publications that draw on the data we have supplied.