Hong Life CalculatorsUnderstanding the sequencing coverage calculator — concept and principle
Sequencing coverage (depth) is the average number of times each base of a genome is read. This calculator estimates mean coverage from genome size, number of reads and read length, and uses the Lander-Waterman model to estimate the fraction of the genome covered at least once. Use it to plan how much sequencing to order, or to check whether a run delivered enough data.
How it is calculated
- Mean coverage: C = N × L × E ÷ G, where N is the number of reads (entered in millions), L the read length in bp, E = 2 for paired-end and 1 for single-end, and G the genome size (entered in Mb).
- Fraction covered ≥ 1×: 1 − e^(−C).
With paired-end selected, the calculator doubles the read count, so enter the number of read pairs (clusters or fragments). If your number already counts both reads, choose "No".
The principle
The total number of bases sequenced is N × L × E. Spreading those bases evenly over a genome of G bases gives the average depth C. Lander and Waterman showed that if reads start at random positions, the number of reads covering any given base follows a Poisson distribution with mean C. The probability that a base is covered zero times is then e^(−C), so the covered fraction is 1 − e^(−C). This falls off quickly: 1× mean coverage covers only about 63% of the genome, while 5× covers about 99.3%.
Worked example
- Inputs: G = 3,100 Mb, N = 600 million read pairs, L = 150 bp, paired-end.
- Bases sequenced: 600 × 10⁶ × 150 × 2 = 1.8 × 10¹¹ bp (180 Gb).
- C = 1.8 × 10¹¹ ÷ 3.1 × 10⁹ = 58.06×.
- Uncovered fraction e^(−58.06) is effectively zero, so the fraction covered ≥ 1× is 100%.
| Genome size (Mb) | Mean coverage (×) |
|---|---|
| 775 | 232.2580645 |
| 1,550 | 116.1290323 |
| 2,325 | 77.41935484 |
| 3,100 | 58.06451613 |
| 3,875 | 46.4516129 |
| 4,650 | 38.70967742 |
| 6,200 | 29.03225806 |
How to read the result
Mean depth is the planning number, but what matters is depth at each position. For human whole-genome germline variant calling, about 30× is a widely used target; somatic tumour studies and low-frequency variant detection typically go much deeper. A theoretical 100% covered does not mean every base is readable: repetitive regions, GC-rich or GC-poor stretches and mapping failures leave gaps in practice.
Common mistakes
- Counting a read pair as two reads and also selecting paired-end, which doubles coverage twice.
- Ignoring duplicates, adapter trimming and unmapped reads, which reduce usable depth.
- Using raw genome size for exome or panel sequencing; use the target region size instead.
Frequently asked questions
How many reads do I need for 30× human coverage?
30 × 3.1 × 10⁹ ÷ (150 × 2) ≈ 310 million read pairs of 2 × 150 bp, before losses.
Is coverage the same as depth?
They are often used interchangeably; "breadth of coverage" refers to the fraction of bases covered.
Why does overlapping paired-end reads matter?
If the fragment is shorter than twice the read length, the two reads overlap and the same bases are counted twice, inflating apparent depth.
Sources: E. S. Lander and M. S. Waterman, "Genomic mapping by fingerprinting random clones", Genomics 2, 231–239 (1988) · D. Sims et al., "Sequencing depth and coverage: key considerations in genomic analyses", Nature Reviews Genetics 15, 121–132 (2014)