Genome Lab · Genomics in your browser
Watch a gene become a protein, then break it
A working 3D model of the machinery inside your cells, driven by real human genes. Every base, codon and amino acid you see is the true sequence. Add a mutation and the machine builds the faulty protein in front of you, then the lab tells you what the change means and designs CRISPR guides to correct it.
Watch it work
39 seconds · the sickle-cell mutation in the haemoglobin geneTry it yourself
What this variant does
Protein, normal vs variant
CRISPR guides to correct it
SpCas9 guides (20 bases + NGG PAM) that cut within 30 bases of the variant, designed on the variant sequence. A variant-specific guide only matches the faulty copy, so it leaves the healthy copy and the repaired gene alone.
| # | Guide (5′→3′) · PAM | Strand | Cut distance | GC | Score | Notes |
|---|
Repair template
Gene at a glance
How it works
Transcription
RNA polymerase unwinds the DNA and copies one strand, base by base, into messenger RNA (T becomes U). About 8 bases of new RNA stay paired to the DNA before it peels away.
Translation
The ribosome reads the mRNA three bases (one codon) at a time. A tRNA whose anticodon matches each codon brings the right amino acid, which is joined to the growing chain.
Folding and release
The chain leaves through a tunnel in the large subunit and starts folding. At a stop codon a release factor frees the finished protein.
What a variant does
The lab rebuilds the protein from the changed DNA, names the change in standard HGVS notation, checks whether the message would be destroyed, and designs CRISPR guides to correct it.
The science behind the numbers
- Impact follows the Ensembl VEP categories: frameshift, nonsense, start-lost and stop-lost changes are HIGH; missense and in-frame changes are MODERATE; synonymous ones are LOW.
- Grantham distance (0–215) measures how different two amino acids are in composition, polarity and size; above 150 counts as radical. BLOSUM62 shows how often a swap is seen between related proteins in nature; negative means rare.
- Nonsense-mediated decay uses the 50–55 base rule: if a premature stop sits more than about 50–55 bases before the last exon junction, the cell usually destroys the message, so little shortened protein is made. Exon boundaries come from the RefSeq records.
- CRISPR scores are simple rules: GC content of 40–70%, no TTTT run (it stops the U6 promoter), a cut within about 10 bases of the variant (repair works best close to the cut), and a bonus when the variant sits in the PAM or seed. Guides that would span an exon junction are left out, because that sequence does not exist in genomic DNA. Off-targets are only checked within this gene: use CRISPOR for a genome-wide search.
What this can and can't do
- It is for learning and research ideas, not diagnosis. Whether a variant causes disease depends on much more than the protein change: check ClinVar and talk to a clinical geneticist.
- The 3D machine is a stylised picture: it shows transcription and translation together (as in bacteria), not to scale, and the folded protein is a sketch, not the real structure. The sequence it reads, the codons and the amino acids are exact.
- Only coding variants are supported, on the main transcript of each gene. Splicing, regulation and the 5′ untranslated region are not modelled.
Sequences: NCBI RefSeq NM_000518.5 (HBB), NM_000492.4 (CFTR), NM_000546.6 (TP53) and NM_000207.3 (INS). Rendering by three.js. Everything runs in your browser; nothing you paste is uploaded.