341: Introduction to Bioinformatics

Dr. Nataa Prulj

Department of Computing
Imperial College London
[email protected]
Winter 2011

Topics
Introduction to biology (cell, DNA, RNA, genes, proteins)
Sequencing and genomics (sequencing technology, sequence
alignment algorithms)
Functional genomics and microarray analysis (array technology,
statistics, clustering and classification)
Introduction to biological networks
Introduction to graph theory
Network properties
Network/node centralities
Network motifs

Network models
Network/node clustering
Network comparison/alignment
Software tools for network analysis
Interplay between topology and biology

Introduction to graph theory

Basic definitions and graph types
Graph representations
Running times of algorithms
Complexity classes
Graph traversing and shortest path problems

Introduction to graph theory

Basic definitions and graph types
Graph representations
Running times of algorithms
Complexity classes
Graph traversing and shortest path problems

Introduction to graph theory

Good textbook (not mandatory):

Chapter 3
Graphs

Slides by Kevin Wayne.

Copyright 2005 Pearson-Addison Wesley.
All rights reserved.

Introduction to graph theory

Graph mathematical object consisting of a set of:
V = nodes (vertices, points).
E = edges (links, arcs) between pairs of nodes.
Denoted by G = (V, E).
Captures pairwise relationship between objects.
Graph size parameters: n = |V|, m = |E|.

Introduction to graph theory

Begun in 1735
Bridges of Konigsberg (todays Kaliningrad):
walk all 7 bridges without crossing a bridge twice
1
3

2
4

6
5
7

Solved by: parity of nodes

Introduction to graph theory

Graph mathematical object consisting of a set of:
V = nodes (vertices, points).
E = edges (links, arcs) between pairs of nodes.
Denoted by G = (V, E).
Captures pairwise relationship between objects.
Graph size parameters: n = |V|, m = |E|.
V = { 1, 2, 3, 4, 5, 6, 7, 8 }
E = { {1,2}, {1,3}, {2,3}, {2,4}, {2,5}, {3,5}, {3,7}, {3,8}, {4,5}, {5,6} }
n=8
m = 11
8

Introduction to graph theory

For graph G(V,E):
V is also denoted by V(G) and VG
E is also denoted by E(G) and EG ; E V x V
Know set operators:

A = B equal sets
A B union of sets A and B
A B intersection of sets A and B
a A a is an element of set A
V = { 1, 2, 3, 4, 5, 6, 7, 8 }
E = { {1,2}, {1,3}, {2,3}, {2,4}, {2,5}, {3,5}, {3,7}, {3,8}, {4,5}, {5,6} }
n=8
m = 11
9

Introduction to graph theory

For graph G(V,E):
If edge e={u,v} E(G), we say that u and v are adjacent or
neigbors
u and v are incident with e
u and v are end-vertices of e
An edge where the two end vertices are the same is called a loop, or
a self-loop

V = { 1, 2, 3, 4, 5, 6, 7, 8 }
E = { {1,2}, {1,3}, {2,3}, {2,4}, {2,5}, {3,5}, {3,7}, {3,8}, {4,5}, {5,6} }
n=8
m = 11
10

Introduction to graph theory

Edge types:
Undirected;

London

Paris

E.g., distance between two cities, PPIs, friendships

Directed; ordered pairs of nodes.

E.g. metabolic reactions, transcriptional regulation,

Loops; usually we assume no loops.

Graph types:
Undirected
Directed
Mixed (some edges
directed some undirected)
Weighted
(weights on edges or nodes)

11

Introduction to graph theory

Undirected edges have end-vertices
Directed edges have a source (head, origin) and target
(tail, destination) vertices
Multigraphs contain multiple edges
Simple graphs
Undirected
No loops or multiple edges

Hypergraphs: E 2V (all subsets of elements of V);

i.e., each edge (hyperedge) is a subset of vertices.
E.g.
V ={v1, v2, v3, v4, v5, v6, v7}
E ={e1, e2, e3, e4}=
={{v1, v2, v3}, {v2, v3}, {v3, v5, v6}, {v4}}

12

Introduction to graph theory

Hypergraphs: E 2V (all subsets of elements of V);
i.e., each edge (hyperedge) is a subset of vertices.
E.g.
V ={v1, v2, v3, v4, v5, v6, v7}
E ={e1, e2, e3, e4}=
={{v1, v2, v3}, {v2, v3}, {v3, v5, v6}, {v4}}

Not used much in network biology, but could be

E.g., metabolic networks where several substances react
with each other to build other substances

13

Introduction to graph theory

Def. An undirected graph G = (V, E) is bipartite if the nodes can be
colored red or blue such that every edge has one red and one blue end.
Applications:
Stable marriage: men = red, women = blue.
Scheduling: machines = red, jobs = blue.
Metabolic networks: metabolites = blue, enzymes = red.

a bipartite graph
14

Introduction to graph theory

Def. A cycle is a path v1, v2, , vk-1, vk in which v1 = vk, k > 2, and
the first k-1 nodes are all distinct.

cycle C = 1-2-4-5-3-1

Simple cycle:
all vertices and edges are distinct
each edge is preceded and followed by its end-vertices
E.g.: 1-2-3-7-8-3-1 in figure above is not a simple cycle, C above is

Cycles denoted by Ck, where k is the number of nodes in the cycle

C3

C4

C5

15

Introduction to graph theory

Def.
An undirected graph is a tree if it is connected and does not contain
a cycle.
Forest does not contain a cycle (so its a union of trees)
Theorem. (illustration only) Let G be an undirected graph on n nodes.
Any two of the following statements imply the third.
G is connected.
G does not contain a cycle.
G has n-1 edges.

16

Introduction to graph theory

Rooted tree. Given a tree T, choose a root node r and orient
each edge away from r.
Importance: Models hierarchical structure.
root r
parent of v

v (an inner node)

child of v
leaves
a tree

the same tree, rooted at 1

17

Introduction to graph theory

E.g.: Phylogeny trees. Describe evolutionary history of species.

18

Introduction to graph theory

Minimum spanning tree. Given a connected graph G = (V, E)
with real-valued edge weights (costs) ce, an MST is a subset of
the edges T E such that T is a spanning tree (contains all
nodes of G) whose sum of edge weights is minimized.
24

23

6
16

18
5

6
5

11

8
14

10

11

21

T, eT ce = 50

G = (V, E)

Cayley's Theorem. There are nn-2 spanning trees of Kn.

can't solve by brute force

19
19

Introduction to graph theory

MST is fundamental problem with diverse applications.
Network design:
telephone, electrical, hydraulic, TV cable, computer, road
Approximation algorithms for NP-hard problems:
traveling salesperson problem, Steiner tree
Indirect applications:
bottleneck paths
identifying functional modules in weighted gene co-expression networks
predicting subgraph structure of protein complexess from affinity purification
studies
clustering

20

Introduction to graph theory

Greedy Algorithms used to find an MST:
Kruskal's algorithm. Start with T = . Consider edges in ascending order
of cost. Insert edge e in T unless doing so would create a cycle.
Prim's algorithm. Start with some root node s and greedily grow a tree T
from s outward. At each step, add the cheapest edge e to T that has
exactly one endpoint in T.

Remark. Algorithms produce an MST.

21

Introduction to graph theory

Def. A complete graph (clique) is a graph on n nodes with all
possible edges between the nodes.
Denoted by Kk, where k is the number of nodes in the clique

K3

K4

K5

22

Introduction to graph theory

Def. A path in an undirected graph G = (V, E) is a sequence P of
nodes v1, v2, , vk-1, vk with the property that each consecutive
pair vi, vi+1 is joined by an edge in E.
So, nodes can repeat, but edges do not.
Def. A walk: a path in which edges/nodes can be repeated.
Def. A path is simple if all nodes are distinct.
A simple path with n nodes is denoted by Pn.
1
2

1
3

Walk:
{2,3}, {3,1}, {1,2}, {2,3}, {3,4}

1
3

Path:
{3,1}, {1,2}, {2,3}, {3,4}

1
3

Simple Path:
{1,2}, {2,3}, {3,4}

Cycle
23

Introduction to graph theory

Def: Two vertices are connected if there is a path between them.
Def. An undirected graph is connected if for every pair of nodes u and v,
there is a path between u and v.
Def. A connected component of G is a maximal connected subgraph of G.
Def. A subgraph G(V, E) of G(V, E): V V, E E (details later).
Def. Maximal connected subgraph of G: subgraph that is connected and is
not contained in any other subgraph of G.

E.g.

3 connected components
24

Introduction to graph theory

A shortest path between two vertices a path of minimal length.
Length number of edges.
Distance between u and v the length of a shortest path between them
(or if a path does not exist)
Subgraphs: G(V, E) is subgraph of G(V, E): V V, E E
Induced sugraph: E contains all edges from E that exist between
nodes in V, i.e., that have both end-points in V
Partial subgraph: contains some of the edges in E that have both
end-points in V.

Partial subgraph:
3-node path

Induced subgraph:
3-node cycle: C4=K4

25

Introduction to graph theory

Degree of a vertex: the number of edges incident to the vertex (in
undirected graphs)
In-degree and out-degree in directed graphs: the number of edges
coming into / going out of the vertex.
E.g.

Property: vV deg(v) = 2m, where |V|=n, |E|=m.

Proof: each end-vertex is counted twice.

Property: in a simple graph, m n(n-1)/2.

Proof: each vertex has degree at most (n-1).
26

Introduction to graph theory

Graph isomorphism

E.g.: Petersen graph, G:

Are there other ways of drawing G?

Is graph H below the same as G above?

27

Introduction to graph theory

Graph isomorphism

28

Introduction to graph theory

Graph isomorphism

29

Introduction to graph theory

Group: an algebraic structure consisting of a set together with an
operation that combines any two of its elements to form a third element.
Operation must satisfy group axioms: closure, associativity, identity and
invertibility.
E.g. of a group: Integers with addition operation: (Z,+).

(not needed, if satisfied, commutative group)

The composition (the application of one function to the results of another)

of two automorphisms is another automorphism
The set of automorphisms of a given graph, under the composition
operation, forms a group, the automorphism group of the graph
30

Introduction to graph theory

E.g.:
node a can be mapped to c by an automorphism, while node b can
only be mapped to itself.
Thus: Orb(a)={a,c}, Orb(b)={b}.
a
b
c
E.g.: in the graphs below, colors denote automorphism orbits.

31

Introduction to graph theory

Basic definitions and graph types
Graph representations
Running times of algorithms
Complexity classes
Graph traversing and shortest path problems

32

Graph Representation: Adjacency Matrix

Adjacency matrix. n-by-n matrix with Auv = 1 if (u, v) is an edge.
Two representations of each edge (symmetric matrix for undirected
graphs; not for directed graphs).
Space: proportional to n2.
Not efficient for sparse graphs (small number of edges compared to
the maximum possible number of edges in the graph),
e.g., biological networks
Algorithms might have longer running time if this representation used
Checking if (u, v) is an edge takes (1) time.
Identifying all edges takes (n2) time.
1
2
3
4
5
6
7
8

1
0
1
1
0
0
0
0
0

2
1
0
1
1
1
0
0
0

3
1
1
0
0
1
0
1
1

4
0
1
0
1
1
0
0
0

5
0
1
1
1
0
1
0
0

6
0
0
0
0
1
0
0
0

7
0
0
1
0
0
0
0
1

8
0
0
1
0
0
0
1
0

= degree of node 2

33

Graph Representation: Adjacency List

Adjacency list. Node indexed array of lists.

Two representations of each edge.

degree = number of neighbors of u
Space proportional to m + n.
Checking if (u, v) is an edge takes O(deg(u)) time.
Identifying all edges takes (m+n) time = linear time for G(V,E).
Requires O(m+n) space. Good for dealing with sparse graphs.

34

Introduction to graph theory

Basic definitions and graph types
Graph representations
Running times of algorithms
Complexity classes
Graph traversing and shortest path problems

35

Running Times of Algorithms

Algorithm:
A sequence of computational steps that transform the input into the
output
A tool for solving a well-specified computational problem
Examples:
BLAST, Smith-Waterman,
Analyzing an algorithm:
Predicting the resources (computational time and memory space)
required
The running time of an algorithm: the number of steps executed as a
function of its input (for G, input is V and E)
Worst-case running time: the longest running time (an upper bound) for
any input of size n.
Order of growth: consider only the leading term of the running time, since
lower-order terms are relatively insignificant for large n.
E.g., O(an2 + bn + c) is O(n2)
An algorithm is efficient if its running time is low order polynomial
36
(quadratic)

Running Times of Algorithms

Why It Matters?

linear

quadratic

cubic
37

Asymptotic Order of Growth

Upper bounds. f(n) is O(g(n)) if there exist constants c > 0 and
n0 0 such that for all n n0 we have f(n) c g(n).

Lower bounds. f(n) is (g(n)) if there exist constants c > 0 and

n0 0 such that for all n n0 we have f(n) c g(n).
Tight bounds. f(n) is (g(n)) if f(n) is both O(g(n)) and (g(n)).
Ex: f(n) = 32n2 + 17n + 32.
f(n) is O(n2), O(n3), (n2), (n), and (n2) .
f(n) is not O(n), (n3), (n), or (n3).

38

Introduction to graph theory

Basic definitions and graph types
Graph representations
Running times of algorithms
Complexity classes
Graph traversing and shortest path problems

39

Complexity Classes
Polynomial-time algorithms:
On input size n, their running time is O(nk)
Not all problems can be solved in polynomial time (poly-time).
Intuition:
Polynomial time algorithms are tractable or easy
Problems that require super-polynomial time are hard
Complexity classes:
P: problems that are solvable in polynomial time
NP: their solutions are verifiable in polynomial time, i.e.,
decision problems for which there exists a
poly-time certifier

Remark. NP stands for nondeterministic polynomial time.

40

Complexity Classes
E.g.
Hamiltonian Cycle of a graph G(V,E) is a simple cycle that contains
each vertex in V.
Problem: does a graph have a Hamiltonian Cycle? NP-complete
If solution given, sequence (v1, v2, v3,,vn) easy to check in poly-time
whether each vi,vi+1 in E for all i and vn,v1 in E .

instance s

certificate t

41

Complexity Classes: P, NP
P. Decision problems for which there is a poly-time algorithm.
NP. Decision problems for which there is a poly-time certifier.
Claim. P NP.
Proof. Consider any problem X in P.
By definition, there exists a poly-time algorithm A that solves X.
If we can solve in poly-time, we can verify a solution in poly time.

Does P = NP? [Cook 1971, Edmonds, Levin, Yablonski, Gdel]

Is the decision problem as easy as the certification problem?
would break RSA cryptography
Clay $1 million prize.
(and potentially collapse economy)

If yes: Efficient algorithms for TSP, FACTOR, SAT,

If no: No efficient algorithms possible for TSP, SAT,

NP

Consensus opinion on P = NP? Probably no.

42

Complexity Classes: P, NP
NP-complete. A problem Y in NP with the property that for every problem X
in NP, X p Y (X is poly-time reducible to Y).
A is poly-time reducible to B if there exists a function f: A B such that is
a yes instance for A if and only if f() is a yes instance for B and if f is polytime computable.
Problem L is NP-complete if:
L is in NP
every problem in NP is poly-time reducible to L (i.e., L is NP-hard:
at least as hard as any NP problem)
NP-hard

NP-c

NP
P

43

Introduction to graph theory

Basic definitions and graph types
Graph representations
Running times of algorithms
Complexity classes
Graph traversing and shortest path problems

44

Graph Traversing
Given a graph G(V,E), explore every vertex and every edge
Using adjacency list is more efficient
Example algorithms:
Depth-first search (DFS)
Breadth-first search (BFS)

45

Graph Traversing
BFS example:

L0
L1

L2

L3
46

Graph Traversing
BFS: code from LEDA
The LEDA Platform of Combinatorial and Geometric Computing, by K.
Mehlhorn and St. Nher

Running time of BFS: linear, O(|V|+|E|), using adjacency list

47

Graph Traversing
DFS applications:
Determines whether G is connected
Computes the connected components of G (strongly connected
components of a digraph = directed graph)
Path / cycle finding
Topological sort (ordering of vertices of digraph G(V,E) such that for every
edge (u,v) in E, u appears before v in the ordering)
Linear running time
BFS applications:
Computes the distance from s to each reachable vertex in unewighted G
Finds shortest paths from s to all other nodes in unweighted G
Finds a simple cycle, if there is one
Computes the connected components of G
48

Graph Traversing
Single-source shortest path problems (SSSPP):
Given a source vertex s, find distances and shortest paths from s to all
other vertices
BFS works on unweighted graphs
Dijkstras algorithm for weighted graphs:
Each node is labeled with its distance from the source node along the
best known path
Initially, all nodes are labeled with infinity

As the algorithm proceeds, labels may change

Label can be:
Non-permanent
Permanent

Initially, all labels are non-permanent

When label represents the shortest possible path from the source to
that node, make it permanent
49

Graph Traversing
Dijkstras Shortest Path Algorithm
Step 1: initially all nodes are non-permament
Step 2: set the source node (A) as permanent
A is at the same time the current node
Step 3:
Examine all non-permanent nodes i adjacent to the current node
Fore each i, calculate the cumulative distance from the source node to i
via the current node
Relabel i with the newly computed distance
But if i already has a shorter cumulative distance than the calculated one, then
to NOT relabel.

Also, label i with the name of the current node (as predecessor)
Compare labels (distances) of all non-permanent nodes and choose the
one with the smallest value. Change the node to permanent and set it as
the current node.
Repeat step 3 until all nodes become permanent.
50

Graph Traversing

51

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

52

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

53

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

54

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

55

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

56

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

57

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

58

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

59

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

60

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

61

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

62

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

63

Graph Traversing
Dijkstras Shortest Path Algorithm
Example

64

Graph Traversing
Dijkstras Shortest Path Algorithm (SSSPP)
Does not allow negative weights on edges
Similar to BFS (BFS is this algorithm but with all weights equal to 1)
Time complexity varies on implementation:
O(|V|2) this is O(|E|) for dense graphs
O(|E| log|V|) good for sparse graphs (for them, O(|E|) is of O(|V|)
O(|V| log|V| + |E|) good for both sparse and dense graphs

65

Graph Traversing
Belman-Ford Algorithm for SSSPP:
Works on weighted, directed graphs
Allows negative weights on edges, but no negative weight cycles
If there is a negative weight cycle reachable from source vertex s, it reports no
solution exists; otherwise produces the shortest paths and their weights
B-F algorithm (G,s)
For each v V
d[v] =
d[s] = 0
For i=1 to |V|-1
For each edge (u,v) E
If ( d[v] > d[u]+w(u,v) )
d[v] = d[u] + w(u,v)
For each edge (u,v) E
If d[v] > d[u]+w(u,v)
Return FALSE (negative weight cycle found)

O(|V|)

O(|V||E|) total running time

- slower than Dijkstras O(|E| log|V|)
for both dense and sparse graphs

O(|E|)
66

Graph Traversing
All-pairs shortest paths
Goal: create an n x n matrix of distance (u,v)
Use B-F algorithm once from each vertex as a source
But O(|V|2 |E|) running time, i.e., O (|V|4) running time for dense graphs
Can do slightly better with Dijkstras from each node, but no negative weight edges
Use adjacency matrix representation of G with entries being weights of edges, wij
Negative weights are allowed, but no negative weight cycles (detects them if exist)
Floyd-Warshall algorithm
Output:
A matrix of distances, D (or equivalently, costs, C)
A predecessor matrix,
Dynamic programming algorithm (breaking down into smaller subproblems)

67

Graph Traversing
All-pairs shortest paths
Floyd-Warshall algorithm
For i=1 to n
For j=1 to n
d(0)ij=wij (d(k-1)ij is length of the shortest i,j-path using only {1,2,,k} nodes)

For k=1 to n
For i=1 to n
For j=1 to n
d(k)ij = min{d(k-1)ij , d(k-1)ik + d(k-1)kj }
Return D(n) (matrix of distances, or costs C)

O(n3) time

O(n2) space
(store only previous matrix)

68

Graph Traversing
Example
Floyd-Warshall algorithm

69

Topics
Introduction to biology (cell, DNA, RNA, genes, proteins)
Sequencing and genomics (sequencing technology, sequence
alignment algorithms)
Functional genomics and microarray analysis (array technology,
statistics, clustering and classification)
Introduction to biological networks
Introduction to graph theory
Network properties
Network/node centralities
Network motifs

Network models
Network/node clustering
Network comparison/alignment
Software tools for network analysis
Interplay between topology and biology

70