3D De Bruijn Genome Assembly Graph

De Bruijn Assembly Graph

k-mer size (k) 5

Coverage (reads) 10

Sequencing error rate 2.0%

Nodes (k−1-mers)

–

Edges (k-mers)

–

Bubbles/tips

–

Assembly

–

How it works

A de Bruijn assembler never aligns whole reads against each other. Instead every read is chopped into overlapping k-length substrings (k-mers); each k-mer becomes a directed edge from its first k−1 letters to its last k−1 letters. Stitching every read's k-mers together this way builds one graph for the whole dataset, and reconstructing the genome becomes graph theory: find a walk that uses every edge exactly once (an Eulerian path). A base miscalled by the sequencer creates a k-mer nobody else supports — a short dead-end "tip" or a parallel "bubble" the path must route around.

nodes  = distinct (k−1)-mers observed
edges  = k-mers (node[i..i+k-2] → node[i+1..i+k-1])
Eulerian path exists ⇔ graph connected AND
  every node has in-degree = out-degree, except
  ≤1 node with out−in=+1 (start) and ≤1 with in−out=+1 (end)

k-mer size (k) — larger k carries more sequence per edge and resolves short repeats, but needs deeper coverage to bridge every join.
Coverage — how many overlapping reads feed the graph; low coverage leaves nodes with no edge out, breaking the path.
Sequencing error rate — each miscalled base mints a spurious k-mer, which shows up as a red tip or bubble the traversal must skip past.
Assembler path — toggle the pulse that walks the highest-confidence route through the graph, the way a real assembler traces contigs out of a de Bruijn graph.

Real-world relevance: de Bruijn graphs are exactly how short-read assemblers (Velvet, SPAdes, most Illumina pipelines) reconstruct genomes at scale — the "many small fragments" outcome real assemblies hit almost always traces back to unresolved bubbles and tips just like the ones visualized here.

Confirmed node Error tip / bubble Traversal pulse

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.