DNA de Bruijn Graph

Bioinformatics Lectures 02
DNA de Bruijn Graph
©Kurt Stüber
31.8.2016
Bioinformatics Lectures 02 Kurt Stueber
A given DNA sequence has two

repeats:
AAGCTTGAGCCgcttGAGA
The DNA sequence is dissected into oligomers of

length 5
AAGCTTGAGCCgcttGAGA
AAGCT GCCGC
AGCTT CCGCT
GCTTG CGCTT
CTTGA GCTTG
TTGAG CTTGA
TGAGC TTGAG
GAGCC TGAGA
AGCCG
There are 12 different oligomers and 3 of them

are found twice
2
AAGCT CGCTT GCTTG
2
AGCCG CTTGA TGAGA
AGCTT GAGCC TGAGC

2
CCGCT GCCGC TTGAG
The oligomers can be linked by 4 base overlaps,

resulting in a looped graph, the de Bruijn graph:
AAGC AAGCT AGCT AGCTT GCTT
CGCT k-mers CTTG

CCGC
CTTGA CTTGA
GCCGC
Overlaps
GCCG TTGA
TTGAG TTGAG
AGCC GAGCC GAGC
TGAG
GAGA
Simplifications:
1. remove k-mers
AAGC AGCT GCTT
CGCT CTTG
CCGC
Overlaps
GCCG TTGA
AGCC GAGC
TGAG
GAGA
Simplifications:
2. fuse unambigous overlaps
AAGC AGCT GCTT
CGCT CTTG
CCGC
Overlaps
GCCG TTGA
AGCC GAGC
TGAG
GAGA
Simplifications:
AAGCT GCTT
CGCT CTTG
CCGC
GCCG TTGA
AGCC GAGC
TGAG
GAGA
Simplifications:
AAGCT GCTT
CCGCT CTTG
GCCG TTGA
AGCC GAGC
TGAG
GAGA
Simplifications:
AAGCT GCTT
GCCGCT CTTG
TTGA
AGCC GAGC
TGAG
GAGA
Simplifications:
AAGCT GCTT
AGCCGCT CTTG
TTGA
GAGC
TGAG
GAGA
Simplifications:
AAGCT GCTT
GAGCCGCT CTTG
TTGA
TGAG
GAGA
Simplifications:
AAGCT GCTT
GAGCCGCT CTTGA
TGAG
GAGA
Simplifications:
3. multipe branches linking the same nodes can be replaced by
single weighted branches.
AAGCT GCTT
2
GAGCCGCT CTTGA
TGAG
GAGA
Simplifications:
1
AAGCT GCTT
2
Start
5 6
GAGCCGCT CTTGA
4 7 3
Stop
8 TGAG
GAGA
Simplifications:
AAGCT GCTTGA
Start
GAGCCGCT
Stop
TGAG
GAGA
Simplifications:
1
AAGCT GCTTGAG
Start
3 2
4
GAGCCGCT
GAGA
Stop
To reconstruct the original sequence find the

eulerian path:
1 4
AAGCT GCTTGAG GAGA
Start Stop
3 2
GAGCCGCT
An eulerian path passes once through every

edge (branch) of the graph.
To reconstruct the original sequence find the

eulerian path:
1 4
AAGCT GCTTGAG GAGA
Start Stop
3 2
GAGCCGCT
=
AAGCTTGAGCCgcttGAGA
Similarly it is possible to build the de Bruijn graph for a sequence

with three repeats.
AAGCTTGAGCCgcttGAGACTTGCTTGAGTAA
GAGACTTGCT
AAGCT GCTTGAG GAGTAA
GAGCCGCT
Two solutions are possible for the reconstruction of the original

sequence: 1-2-3-4-5-6 or 1-4-5-2-3-6, depending on which loop
in the graph is followed first.
GAGACTTGCT
2 3
1 6
5 4
GAGCCGCT
Two solutions are possible for the reconstruction of the original

sequence: 1-2-3-4-5-6 or 1-4-5-2-3-6, depending on which loop
in the graph is followed first.
GAGACTTGCT
2 3
1 6
5 4
GAGCCGCT
=
or
AAGCTTGAGACTTGCTTGAGCCGCTTGAGTAA
The DNA sequence could also be dissected into longer oligomers

for instance of length 10 (k-mer length 10)
AAGCTTGAGCCgcttGAGA
AAGCTTGAGC AGCCGCTTGA
AGCTTGAGCC GCCGCTTGAG
GCTTGAGCCG CCGCTTGAGA
CTTGAGCCGC
TTGAGCCGCT
TGAGCCGCTT
GAGCCGCTTG
Now all oligmers are unique since they are longer than the
repeats and the reconstruction of the original sequence is
straightforward.
AAGCTTGAG AGCTTGAGC GCTTGAGCC
CTTGAGCCG TTGAGCCGC TGAGCCGCT
GAGCCGCTT AGCCGCTTG GCCGCTTGA
CCGCTTGAG CGCTTGAGA
Error correction:
During sequencing the reads produced often have errors, i.e.
wrong bases which have not been correctly assigned.
CTTGAGCCGC TTGAGCCGCT TGAGCCGCTT
TTGAGCCGCA
TGAGCCGCAG
These errors are often found on dead end side branches of the
de Bruijn graph and can be cut off.
CTTGAGCCGC TTGAGCCGCT TGAGCCGCTT
TTGAGCCGCA
TGAGCCGCAG
Summary
De Bruijn graphs can be used to reconstruct (assemble) a DNA sequence

from smaller fragments.
The reconstruction will not always be unique if the sequence contains

repeats longer than the oligomer (or k-mer) length used.
Using longer k-mers helps with the assembly but the length is limited
due to the sequencing technologies.
The graph also helps to find and eliminate errors made during
sequencing.
Links
Wikipedia entry for „de Bruijn graph“:

https://en.wikipedia.org/wiki/De_Bruijn_graph
Python implementation:
https://gist.github.com/BenLangmead/5298132
De Bruijn Graph assembly written by Ben Langmead:
http://www.cs.jhu.edu/~langmea/resources/lecture_notes/a
ssembly_dbg.pdf
Kurt Stueber
Exercise 1
Complete the non-
1 5 10 simplyfied de Bruijn
AGGTCGTCGAGG graph of the sequence
given. The k-mer length
is 4 bases with 3 bases
overlap.
AGG
AGG
Kurt Stueber
Exercise 2 Reconstruct the original

sequence from this
simplyfied de Bruijn
TAATCGT graph with 4 Bases
overlap.
AGGTCGT TCGTAAT TAATGGC TGGCTTT
TGGCTAAT

DNA de Bruijn Graph

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DNA de Bruijn Graph

Uploaded by

Copyright:

Available Formats

Bioinformatics Lectures 02

DNA de Bruijn Graph

A given DNA sequence has two

The DNA sequence is dissected into oligomers of

There are 12 different oligomers and 3 of them

AGCTT GAGCC TGAGC

The oligomers can be linked by 4 base overlaps,

AAGC AAGCT AGCT AGCTT GCTT

CGCT k-mers CTTG

AAGC AGCT GCTT

AAGC AGCT GCTT

To reconstruct the original sequence find the

An eulerian path passes once through every

To reconstruct the original sequence find the

Similarly it is possible to build the de Bruijn graph for a sequence

AAGCT GCTTGAG GAGTAA

Two solutions are possible for the reconstruction of the original

Two solutions are possible for the reconstruction of the original

The DNA sequence could also be dissected into longer oligomers

AAGCTTGAG AGCTTGAGC GCTTGAGCC

CTTGAGCCG TTGAGCCGC TGAGCCGCT

GAGCCGCTT AGCCGCTTG GCCGCTTGA

CTTGAGCCGC TTGAGCCGCT TGAGCCGCTT

CTTGAGCCGC TTGAGCCGCT TGAGCCGCTT

De Bruijn graphs can be used to reconstruct (assemble) a DNA sequence

The reconstruction will not always be unique if the sequence contains

Wikipedia entry for „de Bruijn graph“:

Exercise 2 Reconstruct the original

AGGTCGT TCGTAAT TAATGGC TGGCTTT

You might also like