Using Bayesian Networks to Analyze Expression Data

1
Using Bayesian Networks to Analyze Expression Data
Nir Friedman Iftach Nachman Dana Peer Institute
of Computer Science, Hebrew University

• Presenter Chai Xiaoyong

2
Outline

• Biological Background
• Bayesian Networks
• Analyzing Expression data
• Technical Aspects
• Experiment
• Conclusion and Future Work

3
Part I Biological Background

• DNA
• Gene

• DNA is a double-stranded molecule
• Hereditary information is encoded
• Complementation rules

• Gene is a segment of DNA
• Contain the information required
• to make a protein

4
Part I Biological Background

• Gene Expression refers to the processes involved
  in converting genetic information from a DNA
  sequence into an amino acid sequence, or protein.
• The processes

(gene)
5
Part I Biological Background

• Transcriptionthe focus of this research

the transfer of the genetic information from DNA
to messenger RNA (mRNA), a complementary copy of
the gene.
6
Part I Biological Background

• Each gene encodes a protein and proteins are the
  functional units of life

• Every gene is present in every cell, but only a
  fraction of the genes are expressed at any time

• Many diseases result from the interaction
  between genes

• Understanding the mechanisms that determine
  which genes are expressed, and when they are
  expressed, is the key to the development of new
  treatments of diseases

7
Part I Biological Background

• Why some genes are expressed while others not, or
  having different expression level?

• gene expressions are not independent

• interactions between genes exist

e.g. the expression of gene A promotes the
expression of gene B
e.g. the expression of gene C inhibits the
expression of gene D

• interactions can be complicated

8
Part I Biological Background

• Traditionally experimental means
• ?inefficient and
  insufficient
• Newly developed technique ?DNA Microarray

Make possible measure and compare quantatively
the expression level of tens of thousands of
genes in cells in a single experiment.
9
Part II Bayesian Networks

• Prior work Clustering of expression data

• Groups together genes with similar expression
  pattern

• Disadvantage does not reveal structural
  relations between genes

• Big challenge

• Extract meaningful information from the
  expression data

• Discover interactions between genes based on the
  measurements

10
Part II Bayesian Networks
11
Part II Bayesian Networks

• A Bayesian Network (BN) is a graphical
  representation of a probability distribution

• Compact intuitive representation

• Useful for describing processes composed of
  locally interacting components

• Have a good statistical foundation

• Efficient model learning algorithm

• Capture causal relationships

• Deals with noisy data

12
Part II Bayesian Networks

• Why is it suitable for this problem?

• Gene expression is an inherently stochastic
  phenomenon

• To capture the nature of interactions between
  genes especially the
  causal connection

A

• Microarray techniques are associated with
  missing and noisy data values

B
13
Part III Analyzing Expression Data

• Practical problem Small data sets

• variables hundreds of or thousands of genes

• samples just tens of microarray experiments

• On the positive side, genetic regulation networks
  are sparse!!!

• Characterize and learn features that are common
  to most of these networks

14
Part III Analyzing Expression Data

• The first feature Markov relations

• Symmetric relation Y is in Xs Markov blanket
  iff there is either an edge between them, or
  both are parents of another variable (Pearl 98).

• Biological interpretation a Markov relation
  indicates that the two genes are related in some
  joint biological interaction or process

15
Part III Analyzing Expression Data

• The second feature order relations

• Global property A is an ancestor of B in all
  the equivalent Bayesian networks learned

• Biological interpretation an order relation
  indicates that the transcription of one gene is a
  direct cause of the transcription of another gene

A
B
16
Part IV Technical Aspects

• Learning algorithm induce network structure
• Sparse Candidate Algorithm.
• Feature estimate extract useful features
• A Bootstrap Approach.

17
Part IV Technical Aspects

• Sparse Candidate Algorithm

• An heuristic, iterative approach

• Identify a relatively small number of candidate
  parents for each variable (gene) based on simple
  local statistics at each iteration (Cin)

PaGn(Xi) Cin
Score(Xi , PaGn-1(Xi)?Xj D ) Score(Xi ,
PaGn-1(Xi) D)
18
Part IV Technical Aspects

• Bootstrap method

• Generate perturbed versions of original data
  set, and learn from them

• For i1 m

Resample with replacement N instances from D
(Di)
Learn on Di to induce a network structure Gi

• For each feature f of interest calculate

conf(f) Si1mf(Gi)/m f(Gi) 1 if f is a feature
in Gi
19
Part V Experiment

• Induce Bayesian Networks for 250 yeast genes from
  76 Microarray measurements

• Analyze features in the networks

20
Part V Experiment

• The map left is an example of Markov relation
  features for gene SVS1.
• The width of edges corresponds to the confidence.

21
Part V Experiment

• List of top Markov relation

22
Part VI Conclusions

• Biological motivation
• The develop of microarray technology asks for
  methodologies that are both statistically sound
  and computationally tractable for analyzing data
  sets and inferring biological interactions from
  them
• Advantages of Bayesian Network models
• Can describe local interaction components
• Can Reveal the structure of the transcription
  regulation process
• Provide clear methodologies for learning from
• Can Deal with uncompleted data sets

23
Part VI Future Work

• Incorporate biological knowledge as a prior
• Model the condition attributes into the network,
  such as temporal indicators, background variable
  and exogenous cellular conditions, etc.
• Learn from continuous data
• Combine Bayesian methods with clustering
  algorithms to learn models over clustered genes

24
VII Some Useful Reference

• A Brief Introduction to Graphical Models and
  Bayesian Networks
• DNA Microarray
• Project description

• http//www.ai.mit.edu/murphyk/Bayes/Bayes.html

• http//www.gene-chips.com/

• http//genome-www.stanford.edu/cellcycle/

25
The End

• Thank You !

26
Complementation rules
A(adanine) C(cytosine) T(thymine)
G(guanine)
C
T
C
A
A
T
T
G
A
G
C
G
27
DNA Microarray (informal, intuitive)

• Testing objects

(gene)
experimental
controlled
Referential cDNA
28
DNA Microarray (conti)

• Microarray plate

each slot corresponds to one gene
all genes to be studied are present