Understanding Genetic Variants: SNPs, Indels, and CNVs
The main categories of DNA variation, why they matter clinically, and how they show up in a VCF file.
Every human genome differs from the reference genome at millions of positions. Most of these differences are harmless background variation; a small number cause disease. Classifying what kind of variant you're looking at is the first step toward interpreting it.
The main variant types
| Type | What changes | Typical size | Example |
|---|---|---|---|
| SNP (Single Nucleotide Polymorphism) | One base swapped for another | 1 bp | A → G |
| Indel (Insertion/Deletion) | Bases added or removed | 1 bp - ~50 bp | ATG → ATCTG |
| CNV (Copy Number Variant) | A whole region duplicated or deleted | 1 kb - several Mb | 3 copies of a gene instead of 2 |
| Structural Variant (SV) | Larger rearrangements - inversions, translocations | > 50 bp | A segment flipped end-to-end |
SNPs are by far the most common - the human genome has roughly 4-5 million of them relative to the reference, most with no functional effect. Indels are the second most frequent class. CNVs and larger structural variants are rarer individually but can have outsized effects because they can duplicate or delete entire genes.
Coding consequences matter as much as size
A single-base change can be far more consequential than its size suggests, depending on where it falls and how it shifts the reading frame:
- Synonymous - changes a codon but not the amino acid it encodes (usually harmless).
- Missense - changes one amino acid for another (effect ranges from harmless to disease-causing).
- Nonsense - creates a premature stop codon, usually truncating and disabling the protein.
- Frameshift - an indel whose length isn't a multiple of 3 shifts every codon downstream of it, typically destroying the protein entirely.
This is why a 1-base insertion can be more damaging than a 300-base in-frame deletion - it's the frame that breaks, not just the size.
From detection to a clinical call
In practice (this is a large part of what I do day-to-day), the workflow looks like:
- Call the variant from sequencing data (see the NGS pipelines note for how).
- Annotate it - what gene, what transcript, what predicted protein effect.
- Classify it using a structured framework - the ACMG/AMP guidelines are the clinical standard, sorting variants into Benign, Likely Benign, VUS (Uncertain Significance), Likely Pathogenic, or Pathogenic based on population frequency, computational predictions, functional studies, and segregation with disease in families.
- Report the result in the context of the patient's phenotype.
Types of Sequence Variants - Duplications, Insertions, Deletions, and Indels