Complete NGS Guide: glossary, planning, depth calculation and assembly
1. Introduction to the NGS workflow
Next-Generation Sequencing (NGS) has revolutionized biological research by enabling high-throughput genomic analysis at an unprecedented scale and cost. However, moving from biological samples to high-quality genomic data requires mastery of bioinformatics terminology, quality controls, and precise experimental design. A solid understanding of the NGS glossary and sequencing depth calculations is essential to avoid under‑sequencing, missing variants, or wasting flow cell capacity. This guide walks you through the entire NGS workflow – from raw reads to assembly and annotation – providing step‑by‑step practical calculations.
2. Raw sequencing data terminology
Reads
A read is the distinct nucleotide sequence produced by the sequencing machine, corresponding to a DNA fragment. "Raw reads" refer to this data before any filtering or processing.
FASTQ format
The standard format for storing both biological sequence and its quality scores. Each read is represented by four lines:
@SEQ_ID GATTTGGGGTTCAAAGCAGTATCGATCAAATAGTAAATCCATTTGTTCAACTCACAGTTT + !''*((((***+))%%%++)(%%%%).1***-+*''))**55CCF>>>>>>CCCCCCC65
Phred quality score (Q score)
Logarithmic measure of base‑call accuracy: Q = -10 log₁₀(P), where P is the error probability. Q20 = 99% accuracy (1 error/100), Q30 = 99.9% (1 error/1000).
Paired‑End (PE) vs Single‑End (SE) reads
- SE: reads from one end of the fragment.
- PE: reads from both ends, generating two reads per fragment. Superior for resolving structural variants and repetitive regions.
3. Quality control (QC) and preprocessing
- Quality distribution: read quality typically drops towards the 3' end.
- GC content: percentage of G+C. Deviation from expected may indicate contamination or preparation bias.
- Adapter trimming: removal of synthetic adapter sequences at the read ends.
- Quality filtering: removal of low‑quality bases (e.g. Q<20).
- Standard tools: FastQC (visualisation), Trimmomatic, Cutadapt, fastp (trimming/filtering).
4. Sequencing depth and coverage
Depth is the average number of times a base is sequenced:
Depth (X) = Total sequenced bases (bp) / Genome size (bp)
Example: 5 Mb genome, 150 Mb data → 30×.
Coverage is the percentage of the genome actually covered by at least one read (e.g. 98%). High depth does not guarantee 100% coverage.
5. Flow cell capacity and platform yields
Illumina platforms
- MiSeq: up to ~15 Gb per run.
- NextSeq 2000: up to ~360 Gb.
- NovaSeq X Plus (S4): up to ~3,000 Gb (3 Tb).
Ion Torrent platforms
- Ion 314 Chip: ~100 Mb.
- Ion 318 Chip: ~2 Gb.
- Ion S5 (540 Chip): ~15 Gb.
6. How to calculate required sequencing data
Steps:
- Determine the target genome size.
- Choose the target depth (e.g. 30× for variants, 50‑100× for de novo assembly).
- Apply the formula:
Required data (bp) = Genome size (bp) × Desired depth
Example: 1.2 Gb genome, depth 50× → 60 Gb data.
7. Calculating number of samples per run
Number of samples = Total flow cell yield / Required data per sample
Example NovaSeq: 300 Gb / 30 Gb = 10 samples.
8. Read mapping terminology
- Reference genome: a digital representative sequence of a species.
- Alignment: assigning each read to its position on the reference genome.
- Mapping rate: percentage of reads successfully aligned (<70% indicates contamination or poor quality).
- SAM/BAM files: standard formats for storing alignments (BAM = compressed binary version).
9. Genome assembly terminology
- De novo assembly: reconstructing the genome without an external template.
- Reference‑guided assembly: uses a closely related genome as a scaffold.
- Contig: a continuous sequence with no gaps.
- Scaffold: a set of contigs linked by paired‑end information.
- N50: the length of the smallest contig at 50% of the total assembly size.
- L50: the number of contigs covering 50% of the assembly.
- BUSCO: a measure of biological completeness based on conserved orthologous genes.
10. Assembly evaluation
- Misassemblies: joining errors, inversions or translocations.
- Coverage consistency: mapping reads back to contigs to validate stability.
- Tools: QUAST (technical metrics), BUSCO (biological verification).
11. Genome annotation terminology
- Structural annotation: identification of genes, exons, introns, promoters.
- Functional annotation: assigning biological functions to predicted genes.
- Open Reading Frames (ORF): potential coding sequences between start and stop codons.
- Gene prediction: tools like Augustus or Prokka.
- Databases: InterPro, Pfam, NCBI RefSeq/GenBank.
Summary table of key terms
| Term | Definition | Formula / Metric | Example |
|---|---|---|---|
| Phred Score (Q) | Base‑call accuracy | Q = -10 log₁₀(P) | Q30 = 99.9% |
| Depth | Average number of reads per base | Total bases / Genome size | 30× |
| Coverage | Percentage of genome covered | % of positions ≥1× | 98% |
| N50 | Median contig length | Shortest contig at 50% | 45 kb |
| L50 | Number of contigs at 50% | Count | 12 contigs |
💬 Commentaires