Umapathy-proposalx - University of South Australia

throneharshΒιοτεχνολογία

2 Οκτ 2013 (πριν από 4 χρόνια και 1 μήνα)

167 εμφανίσεις

1



School of Computer and Information Science

University of South Australia



Thesis Proposal


Discovery of interaction modules between hypoxia regulated mRNA
and

miRNA by Splitting Averaging
-
Bayesian Network Learning



By

Jeya Karthika Pandian Umapathy



Supervisor

Dr. Lin Liu



13
th

June 2010


INFT 4017

CIS Research Methods

Master of Science

(Computer and Information Science)

2



Abstract

Biologists suspect that
the miRNA mediates the hypoxia induced suppression of mRNA
expression post
-
transcriptionally, in breast cancer affected cells. Manual discovery of interaction
modules between the miRNA and mRNA proves to be impossible owing to the large data set.
Hence, we

use a computational method using Splitting Averaging


Bayesian Networks,
proposed by Liu et al., for learning any connections between miRNA and mRNA datasets
derived from breast cancer affected cells. The research helps the biologists in confirming their

assumptions. We also propose to validate this computational method proposed by Liu et al., in
terms of usefulness and effectiveness
,

in discovering the miRNA
-
mRNA interactions.



Keywords


miRNA, mRNA, Hypoxia,
HIF,
gene expression, post
-
transcriptional, down
-
regulation,

Up
-
regulation, mixed
-
regulation, Splitting Averaging Bayesian network strategy.














3


Contents

1.

Introduction








...4

1.1.

Background







…4

1.1.1.

Ribonucleic Acids (RNAs)




…4

1.1.2.

Messenger RNA (mRNA)




…5

1.1.3.

MicroRNA (miRNA)





…5

1.1.4.

miRNA functions






…6

1.2.

Research questions






…6

1.3.

Scope and limitations





…7

2.

Literature Review







…8

2.1.

Why to learn miRNA
-
mRNA interactions?


…8

2.2.

Effect of Hypoxia on mRNA




…8

2.3.

Computational
Methods



Why?




…9

2.4.

Related
Works






…9

3.

Research Methodology






…10

3.1.

SA


BN Strategy






…11

3.2.

Process flow







…12

3.3.

Tools Used







…12

3.4.

Data
Sources







…13

4.

Expected Outcomes







…13

5.

Conclusion








…14


Appendix A


Extended Abstract





…15

Appendix B


Process
flow of the SA
-
BN strategy



…16

Appendix
C



miRNA interactions inside the cell



…17

Appendix D


Time Line for the research




…18

Appendix E


Tentative thesis chapters outline



…21

References








…23




4


1.

Introduction


The advancement in the computing field marked the beginning of Bioinformatics during
1960's. With the advent of internet, the field was blotted with remarkable findings which would
have otherwise been difficult.
The alliance between genomics and bioinforma
tics aided the quest
to reveal the mechanisms going on inside the minuscule cell of the human body at molecular
level.


1.1.

Background

In an attempt to understand the events at molecular level, it was discovered that
ribosomes were the hub of protein synthesis

and that the messages
were

transmitted from DNA
to the ribosomes through RNA (Paustian & Roberts 2006). This discovery by Sydney Brenner,
Francois Jacob and Matthew Meselson in 1961 accelerated the interests in the field and raised
numerous

questions in t
he minds of biologists (Paustian & Roberts 2006).



1.1.1.

Ribonucleic Acids (RNAs)

RNA became the source of attraction to biologists and was investigated frequently. The
RNA's were found to constitute about 60% of the ribosome’s weight and it plays an
indispensable role in catalyzing the protein synthesis (Moore & Steitz 2002).

RNA or
the Ribonucleic Acid is a nucleotide polymer, which is 21 to 25 nucleotides in
length (Lagos
-
Quintana et al. 2001). Each nucleotide has a nitrogenous base, a ribose sugar, and
a phosphate (Barciszewski & Clark 1999). The nitrogenous bases found in RNA are
Cytosine,
Guanine, Adenine and, Uracil
,

which are abbreviated respectively as C, G, A and U.

5


As a part of ribosome, RNA has a biological significance owing to its role in DNA
-
Ribosome communication and protein synthesis (Clancy 2008). RNA has been classif
ied into
numerous types based on its function as snRNA or Small Nuclear RNA, siRNA or Small
Interfering RNA, snoRNA or Small Nucleolar RNA, miRNA or microRNA, tRNA or transfer
RNA, mRNA or messenger RNA and rRNA or the ribosomal RNA (Clancy 2008).


1.1.2.

Messeng
er RNA (mRNA)

The messenger RNA
is a single stranded RNA, which
plays the role of the
communicator.
The mRNA is formed inside the nucleus of the cell by the transcription of DNA
molecule. After the mRNA matures, i
t transports the “protein
-
blue prints” from

the DNA to the
ribosomes
in the
cytoplasm
, for the purpose of protein synthesis

(Clancy 2008 & Johnston &
Bose 1972).

The interactions of the mRNa inside the cell can be understood from the figure in
Appendix C
.


1.1.3.

MicroRNA (miRNA)

The first miRNA, lin
-
4
was initially found to exist in C. elegans (Bentwichet al. 2005,
Berezikov et al. 2006, Lee et al. 1993 & Wightman et al. 1993). From then on it was tracked
through numerous species including humans and its functions were elucidated. So far many
hundreds o
f miRNAs have been found and this figure is speculated to increase tremendously
with further investigation of cells at molecular level (He & Hannon 2004). It has been found that
they are mostly single stranded and non
-
coding nucleotide polymers (Lagos
-
Quin
tana et al.
2001).


6


1.1.4.

miRNA functions

The miRNAs play a prominent role in cellular activities
ranging
from cell differentiation
to
development

(Ambros 2004, Bushati & Cohen 2007 & Du & Zamore 2007). The
identification of the regulatory targets of the miRNAs
aids in determining the miRNA functions
based on the functions of the target (Bartel 2004, Enright et al. 2003 & Lewis et al. 2003).

In the words of Barciszewski & Clark (1999, pp. 10), miRNA performs a ‘feat of
gymnastics’ in the cellular level. Changes i
n miRNA mediates changes in the neural system (De
Pietri Tonelli et al. 2008), immune system (Schickel 2008), cardiac system (Divakaran et al.
2008), etc.. The most significant being its ability to control the translation of mRNA either by
accelerating or
degrading its expression post
-
transcriptionally (Shyu et al. 2008).
It can up
-
regulate, down
-
regulate and mix
-
regulate mRNA gene expression.

1.2.

Research questions

The messenger RNA which is affected by Hypoxia (
Hypoxia

is actually a condition of
human body during which the
oxygen supply to the tissues is well below the normal
physiological requirement. This is prevalent in case of an occurrence of certain abnormal
changes in the tissue or cell
)

has been observed to resu
lt in the repression or down
-
regulation of
mRNA

in the MCF7 breast cancer derived cells
. The biologists suspect that this may have been
mediated by the changes occurring in
miRNA. We propose to test this hypothesis by applying a
computational method propos
ed by Liu et al., (2009b) and finding whether miRNA has any
interesting interactions with the
hypoxia affected mRNA.

7




According to the work of Mole et al. (2009), it is found that Hypoxia Inducible Factor
(HIF)
plays an indirect role in the
gene
repression of hypoxia affected mRNA. Hence this raises
a new suspicion that the gene repression by hypoxia could have been mediated by the miRNA.

So, the construction of the network of miRNA
-
mRNA interactions will enable the inference of
any interesti
ng connections between them leading to the confirmation of the hypothesis.



1.3.

Scope and limitations

The thesis
focuses

only
on
the breast cancer affected cells. It focuses on finding the
interactions between the miRNA and hypoxia regulated mRNA, which may prove that the gene
repression of hypoxia regulated mRNA was mediated by the
changes in
miRNA.

The research
may or may not find

new interactions between miRNA and mRNA. However it does not
concentrate on finding any new connections between mRNA and miRNA.

The thesis also focuses on validating the computational method proposed by Liu et al.
(2009b) in terms of the usefulness and e
ffectiveness in this
particular
application area.

mRNA affected by
Hypoxia

?

Down
-
regulation
of mRNA

mRNA affected by
Hypoxia

Mediated by
changes in miRNA

Down
-
regulation
of mRNA

8



2.

Literature Review

2.1.

Why to learn miRNA
-
mRNA interactions?

The
changes in miRNA
gene
regulation are

known to cause abnormalities in human body
leading to diseases
like

cancer (Wienholds & Plasterk 2005).
U
p
-
regulation
,

down
-
regulation

and mixed
-
regulations

were observed
to be caused by

miRNA but mostly, down
-
regulation
dominated in tumors (Lu et al. 2005). The disruption of its rheostatic function (Baek et al. 2008)
causes tumor development and also acceler
ates the carcinogenic tumor growth to malignant
state. The study of miRNA regulation networks and targets, will henceforth aid us in
understanding the cause for the abnormal physiological conditions and enlighten the otherwise
unknown biological procedures

of human body (Liu et al. 2009a).


2.2.

Effect of Hypoxia on mRNA

Hypoxia

can be described as

a condition in
human body during which the whole human
body or merely a part of it
is
affected by a decrease in the
oxygen supply
. This condition helps in
the process

of proper metabolism in the cellular level and also sometimes the regulation of the
numerous genes in the human body (Mole et al. 2009).

The blood flowing through the arteries
delivers the oxygen to the cells by diffusion. During this diffusion, the
partial pressure is usually
100mmHg. But, if this pressure falls below 40mmHg, it becomes lethal. When this insufficiency
occurs, lactic acid is formed from the hydrogen for producing little energy by temporary
anaerobic metabolism. The increase in lactic
acid inside the cell may cause inadequate blood
flow, hypoxemia, etc
..,

often leading to death

(Roach et al. 2001)
.

9


Recent research on the MCF7 cells (breast cancer affected cells) reveals that hypoxia
suppresses the mRNAs from expessing their genes. The r
esearch by Mole et al. (2009) reveals
that the Hypoxia Inducible Factor (HIF) indirectly suppresses the gene regulation. Hence, it
raises the possibility that miRNAs could have
been
mediated the down
-
regulation of mRNA
suppressed by hypoxia.


2.3.

Computationa
l methods

-

Why
?

The biologists obtained the mRNA (Elvidge et al. 2009) and miRNA (Camps et al. 2009)
data from the breast cancer derived cells.
On

the precinct of verifying the hypothesis,
the large
number of possible combinations proves manual analysis a
nd verification
as impossible
. Hence,
it is difficult for the biologists to test every possible connection between miRNA and mRNA
pairs to verify the hypothesis. So, we utilize computational approaches and methods to identify
any connections between the mi
RNA and mRNA data.


2.4.

Related

Works

The
past few years

has seen many computational methods for the purpose of validating
the possible hypotheses regarding the miRNA
targeting

information. In 2005, Yoon et al. came
up with a new computational method based on prediction, depending on the idea that
the
binding
structure between the miRNA and the mRNA will be normal and it’s the same even if many
binding sites exists on the mRNA.

He used weighted bipartite graphs to form the binding
structures between the micro and messenger RNA. But, this resulted in a higher rate of false
discovery since it relied only on the sequence data.

10


Bayesian parameter learning was used by Huang et al.
(2006) to learn the interactions
between miRNA and mRNA using both sequence data and expression data. The bi
-
clustering
approach (Joung et al. 2007) utilizing the expression profile data and sequence information
ventured to find the miRNA regulatory module
s (MRMs). A rule
-
based approach was proposed
by Tran et al., to study miRNA
-
mRNA interactions assuming that expression profiles of miRNA
and mRNA will be quite similar in a given module. The above mentioned methods have minimal
false discovery rate but, fa
il to use a sample category which is a critical factor since many of the
biological experiment data measure up to different phenotypic or conditional groups.

The functional miRNA
-
mRNA regulatory modules (FMRMs) were learnt for the miRNA
and target mRNA sp
ecial conditions by Liu et al. (2009a). But, this work focused only on down
-
regulation. The next paper by Liu et al., (2009b) used Splitting Averaging
-

Bayesian Network
strategy to discover the miRNA
-
mRNA interactions utilizing the target sequencing infor
mation,
expression profiles of miRNA and mRNA. This work cover
s

all
the

down & up
regulation and
also provided a method to analyze miRNA
-
mRNA interactions in various physiological
disorders.

This strategy posed minimal false discovery rate by using the sam
ple category
information. Hence, we propose to use this strategy for learning the miRNA
-
mRNA interactions.
Additionally, due to the fact that the author of the paper works in the same university, we are
able to get a good amount of guidance and information
.


3.

Research
Methodology

Of the various Computational methods,
Construction of the n
etwork structure has gained
importance in building diagnostic models for diseases like cancer

(Sebastiani et al. 2004)
.

And,
11


Bayesian network learning has been found to best

serve the purpose.

Variations in the normal
Bayesian network structure learning yields better results

in our application area
.


3.1.

SA
-
BN Strategy

The methodology employed to
discover the regulatory interaction modules between
hypoxia regulated mRNA and miRNA is the Splitting Averaging
-
Bayesian Network Learning
strategy.

“Bayesian network is actually a probabilistic graphical model which signifies a set of
arbitrary variables
and their conditional dependencies using directed acyclic graph”

(Heckermann 1995)
.

It can be used to construct the probabilistic relationship network for
pathological conditions. A Bayesian network consists of n nodes (representing a random
variable, attr
ibute or a hypothesis) connected via edges (representing the existence and direction
of relationship). It can be used to learn unobserved variables in the network, for learning about
the parameters, or to learn about the structure. There are numerous algor
ithms for learning
Bayesian networks.


In Splitting Averaging


Bayesian network learning, we actually split the dataset based
on the sample category information. This is followed by Bayesian network learning for the split
data. Now, The Bayesian networks
learnt from split data are merged together in to a single
network using the averaging strategy. This reduces the false discovery rate.

The process of learning Bayesian networks computationally proves to be impossible,
owing to the large data set and the fa
ct that exponentially increasing number of network
structures
is

possible. We utilize the miRNA targeting information to limit such possibilities (Liu
12


et al. 2009b). Hence,
we utilize the target sequencing information to minimize the false
discovery rate (
Liu et al., 2009b).


3.2.

Process Flow

In this method, we use the expression profiles of miRNA and mRNA along with the
target sequencing information to learn a Bayesian network of miRNA
-
mRNA interactions.
The
normalized differ
entially expressed profile datasets

of miRNA and mRNA are split based on
sample category information. We discretize the expression profile data as a measure to
standardize them, since it is obtained from different platforms. We now use the Bayesian
network learning to obtain the interaction

dependencies between the discretized data of miRNA
and target mRNA. The two structures learnt from the split data are now merged using the
averaging strategy of Bayesian Networks. This process is illustrated in the process flow diagram
in the Appendix B.


3.3.

Tools Used

The
free
open source data mining tool “R package”

written initially by
Robert Gentleman
and Ross Ihaka

in 1997 and improved till date,
is used
for the purpose of

pre
-
process
ing

the
datasets.
The R package tool is continuously being used for the

statistical purposes and graphs. It
has nearly become the de
-
facto for data
-
mining. We use the current version, R 2.11.

released on
31
st

May 2010 for our research.

Further to R, we also use the Bio
-
conductor package, “BioC” for the purpose of analysis
of
genomic data. This is an open source add
-
on for R package especially to analyse DNA and
13


RNA microarray experiment datasets. We use the current developer version, BioC 2.
6

for our
research.


3.4.

Data Sources

We use heterogeneous data in our computational approa
ch. This includes miRNA target
information, expression profiles of miRNA and expression profiles of mRNA.

Numerous databases are available for the miRNA targeting information, for example,
miR
-
200 can be used for our research
(
Griffths
-
Jones, 2008
).

We use
the
differentially expressed profile data of miRNA obtained from Array Express
(http://www.ebi.ac.uk/microarray
-
as/ae/) using the accession number E
-
MEXP
-
1111
(Camps et
al. 2009)
. The differentially expressed profile data of mRNA is obtained from GE
O
(http://www.ncbi.nlm.nih.gov/geo/) using the GEO accession number GSE3188
(Elvidge et al.
2009).

These are samples from the breast cancer affected

MCF7

cells.


4.

Expected Outcomes

We expect to obtain the possible connections between the hypoxia regulated m
RNA and
changes in miRNA after the final merging of the Bayesian network. We expect to find some of
the targets for miRNAs from the mRNA repressed by Hypoxia.

The results will confirm the
already proved literature and
may
enlighten us with new
possibilities.

The use of heterogeneous
data and sample categories is expected to minimize the false discovery rate.
We also expect to
validate the usefulness and effectiveness of this computational method proposed by Liu et al.,
(2009 b).

14



5.

Conclusion

In t
his research, we utilize an existing computational approach of Splitting Averaging


Bayesian Network learning to infer the complex miRNA
-
mRNA interactions in breast cancer
affected cells. The research helps in validating the hypothesis of the effect of hy
poxia on the
breast cancer derived cells

as well as in

determin
ing

the usefulness of the SA
-
BN strategy in
inferring the miRNA
-
mRNA interactions.















15


A
ppendix A


Extended Abstract


The
R
ibonucleic Acid (RNA
)
, which i
s a nucleotide polymer, plays
a

prominent role in
protein synthesis

and the
regulation of gene expression
.
Of the various types of RNA
s
, the
microRNA (miRNA) controls messenger

RNA (mRNA) translation by accelerating or degrading
its
expression
.
This has been found to be the cause for

v
arious
abnormalities in human bod
ies
.
Current research reveals that miRNAs are responsible for gene regulation
s

by post
-
transcriptional control of their

mRNAs
a
nd, it may result in the development of tumor or
even
inducing the carcinogenic tumor growth to
malignant state. Hence it proves vital to know the
mi
RNA
-
mRNA interactions to aid
the prevention and treatment of various physiological
conditions.


Recent research on the breast cancer affected cells reveals that the expression of mRNAs
is suppressed by
hypoxia. Biologists have discovered that HIF (Hypoxia Inducible Factor)
indirectly regulates the gene expression. This has led them to assume that miRNAs could be the
indirect reason for suppression of

mRNA
expression by hypoxia. However, given the large
n
umber of mRNAs and miRNAs
,

it is impossible for biologists to test each
and every
miRNA
-
mRNA pair, to verify their assumption.


In this research,

we apply a computational
method
proposed by Liu et al., on the miRNA
and mRNA expression data sets obtained
from breast cancer derived cells, to discover possible
connections between the miRNAs and mRNAs and, help biologists to verify their assumption on
the role of miRNAs mediating the suppression of mRNA expressions by hypoxia. At the same
time, we want to val
idate the usefulness and effectiveness of the method proposed by Liu et al.,
in discovering meaningful miRNA
-
mRNA interactions in this application area.





16


Appendix B


Process flow of the S
A
-
BN Strategy




(Source:
Liu, B, Li, J, Tsykin, A, Liu, L, Gaur, A & Goodall, G 2009b, 'Exploring complex
miRNA
-
mRNA interactions with Bayesian networks by splitting
-
averaging strategy',
BMC
Bioinformatics
, vol. 10, no. 1, pp. 408
)

17


Appendix
C



miRNA interactions inside the cell




(Source:
National Institutes of Health, “Talking Glossary of Genetic Terms”, National Human
Genome Research Institute, Viewed on June 12 2010,
<http://www.genome.gov/glossary/?id=123>
)


18


Appendix D


Time Line for Study Period 2, 2010

SP 2, 2010

Tasks

Week 1

Supervisor Search Finalization
.

Meeting the Supervisor
.

Week 2

Decide on the Research Area
.

Go through the base papers
.

Week 3

Start Background Study on Bio
-
informatics
.

Learn about Bayesian Networks
.

Week 4

Decide on meeting schedule and working

space
.

Complete the Induction program
.

Continue background stud
y.

Week 5

Deeper discussion about the research area
.

Determine the exact objectives and requirements of the
research
.

Continue Background study
.

Week 6

Start working on the thesis abstract
.

Submit a document to test the understanding about the
research
.

Discussion of the document and base paper
.

Teaching

Break

Start working on literature review
.

Learn how to use R package
.

Teaching

Break

Submit an annotated list of bibliography to be used
in the
literature review.

19


Discussion of the annotated bibliography.

Completion and review of the thesis abstract
.


Week 7

Submission of thesis abstract.

Start working on thesis proposal
.

Week 8

Work on the literature review
.

Meeting with Bing Liu to
understand the concepts and
methodology to be used
.

Week 9

Complete the literature review.

Start on the methodology and remaining parts of proposal
.

Week 10

First draft
submission
to supervisor
.

Prepare Extended Abstract and send to supervisor.

Week 11

Feedback discussion for the first draft and extended
abstract.

Submit Extended Abstract

Week 12

Second draft submission to supervisor and feedback
.

Third draft
submission
to Supervisor and feedback
.

Presentation slides preparation and feedback
.

Week 13

P
resentation and Thesis Proposal.








20


Time Line for Study Period 5, 2010

SP 5, 2010

Tasks

Week 1

Start Data Preparation.

Week 2

Complete data preparation.

Week 3

Input the data to the Software for learning the miRNA and
mRNA connections and obtain
results.

Week 4

Analyze the
results
.

Work on the thesis Literature review.

Week 5


Work on the
Th
esis Methodology.

Week 6

Finalize the results of analysis
.

Complete the Literature review and Methodology.

Week 7

Validate the methodology used.

Determine
its usefulness and efficiency.

Teaching

Break

Work on the results for the thesis.

Teaching

Break

Work on
recommendations and conclusion of the thesis.

Week 8

First draft
submission
to supervisor
.

Week 9

First draft feedback

and discussion.

Prepare
abstract for the thesis.

Week 10

Second draft
submission
to supervisor
.

Week 11

Second draft feedback

and discussion.

Prepare p
resentation slides
.

Week 12

Final draft of thesis and discussion.

Presentation slides feedback

and discussion

Week 13

Presentation
and

Final Thesis
submission.

21


Appendix
E


Tentative Thesis Chapters

Outline


6.

Introduction

6.1.

Background

6.2.

Motivation

6.3.

Research Purpose

6.4.

Research Objectives

6.5.

Scope and Limitations

6.6.

Thesis Organization


7.

Literature Review

7.1.

Background

7.2.

miRNA and mRNA
Interaction

7.3.

Effect of Hypoxia

7.4.

Methods to discover
miRNA
-
mRNA interaction modules

7.5.

Computational discovery of miRNA
-
mRNA interaction modules

7.6.

Various Computational methods

7.7.

Use of Bayesian Networks in Bio
-
informatics

7.8.

Splitting Averaging


Bayesian Network
Learning

7.9.

Related

Works


8.

Research Methodology

8.1.

Data sources

8.2.

Data mining tools

8.3.

Data Pre
-
processing

8.3.1.

Splitting the data

8.3.2.

Results and observations

8.4.

Learning the interaction Bayesian network

22


8.4.1.

Learning network from the Split dataset

8.4.2.

Combining the network by Averaging

strategy


9.

Results and Proceedings

9.1.

Identifying the c
onnections between miRNA and mRNA

9.2.

Inference from the SA
-
BN learning

9.3.

Observations

9.4.

Validation of SA
-
BN Strategy

9.4.1.

Usefulness

9.4.2.

Effectiveness

9.4.3.

Results and observations


10.

Discussion


11.

Recommendations


12.

Conclusion


13.

References











23


References

Ambros, V 2004, 'The functions of animal microRNAs',
Nature
, vol. 431, no. 7006, pp. 350
-
355.

Baek, D, Villen, J, Shin, C, Camargo, FD, Gygi, SP & Bartel, DP 2008, 'The impact of
microRNAs on protein output',
Nature
, vol. 45
5, no. 7209, pp. 64
-
74.

Barbato, C, Arisi, I, Frizzo, ME, Brandi, R, Da Sacco, L & Masotti A 2009, 'Computational
Challenges in miRNA Target Predictions: To Be or Not to Be a True Target?',
Journal of
Biomedicine & Biotechnology
.

Barciszewski, J & Clark, B
FC 1999, 'Why RNA?' in
RNA Biochemistry and Biotechnology
,
Kluwer Academic Publishers, Netherlands, pp. 1
-
10.

Bartel, DP 2004, 'MicroRNAs: genomics, biogenesis, mechanism, and function',
Cell
, vol. 116,
pp. 281
-

197.

Bentwich, I, Avniel, A, Karov, Y, Aharonov, R, Gilad, S & Barad, O 2005, 'Identification of
hundreds of conserved and nonconserved human micrornas',
Nature Genetics
, vol. 37, pp. 766

770.

Berezikov, E, Cuppen, E & Plasterk, RHA 2006, 'Approaches to microrn
a discovery',
Nature,
Genetic
s,

vol. 38
, pp. S2

7.

Bushati, N & COHEN, S 2007, 'microRNA Functions',
The Annual Review of Cell and
Developmental Biology
, vol. 23, no. 1, pp. 175
-
205.

Camps, C, Buffa, FM, Colella, S, Moore, J, Sotiriou, C, Sheldon, H, Harri
s, AL, Gleadle, JM &
Ragoussis, J 2008,'hsa
-
miR
-
210 is induced by Hypoxia and is an independent prognostic factor
in breast cancer',
American Association for Cancer Research Journal
, vol. 14, no. 5, pp. 1340
-
1348.

Clancy, S 2008, 'RNA Functions',
Nature Ed
ucation
, vol. 1, no. 1.

24


De Pietri Tonelli, D, Pulvers, JN, Haffner, C, Murchison, EP, Hannon, GJ & Huttner, WB 2008,
‘miRNAs are essential for survival and differentiation of newborn neurons but not for expansion
of neural progenitors during early neurogen
esis in the mouse embryonic neocortex.
Development’, vol. 135, pp. 3911

3921.

Divakaran, V, Mann & DL 2008, ‘The emerging role of miRNAs in cardiac remodeling and
heart failure’,
Circulation Research
, vol. 103, pp. 1072

1083.

Du, T & Zamore, PD 2007, 'Beg
inning to understand microRNA function',
Cell
, vol. 17, no. 8,
pp. 661
-
663.

Elvidge, GP, Glenny‡, L, Appelhoff, RJ, Ratcliffe, PJ, Ragoussis, J & Gleadle, JM 2006,
'Concordant Regulation of Gene Expression by Hypoxia and 2
-
Oxoglutarate
-
dependent
Dioxygenas
e Inhibition : the role of HIF
-
1, HIF
-
2, and other pathways',
The Journal of
Biological Chemistry
,

The American Society for Biochemistry and Molecular Biology
, vol. 281,
no. 22, pp. 15215
-
15226.

Enright, AJ, John, B, Gaul, U, Tuschl, T, Sander, C, & Mark
s, DS 2003, 'MicroRNA targets in
Drosophila',
Genome Biology
, vol. 5.

Griffths
-
Jones, S, Saini, HK, van Dongen, S & Enright, AJ 2008, 'miRBase: tools for microRNA
genomics',
Nucleic Acids Research
, vol. 36, no. 1, pp. D154
-
158.

He, L & Hannon, GJ 2004, 'Mi
cro RNAs: Small RNAs with a big role in gene regulation',
Nature Reviews
, Genetics, vol. 5, pp. 522

531.

Heckermann, D 1995, 'A Tutorial on Learning with Bayesian Networks',
Innovations in Bayesian
Networks
, vol. 156, pp. 33
-
82.

25


Huang, JC, Morris, QD & Fre
y, BJ 2006, 'Detecting MicroRNA Targets by Linking Sequence,
MicroRNA and Gene Expression Data',
Research in Computational Molecular Biology
, vol.
3909, pp. 114
-
129.

Johnston, RE & Bose, HR 1972, 'Correlation of Messenger RNA Function with Adenylate
-
Rich
S
egments in the Genomes of Single
-
Stranded RNA Viruses',
Proceedings of the National
Academy of Science
, USA, vol. 69, no. 6, pp. 1514
-
1516.

Joung, JG, Hwang, KB, Nam, JW, Kim, SJ & Zhang, BT 2007, 'Discovery of microRNA mRNA
modules via population
-
based pr
obabilistic learning',
Bioinformatics
, vol. 23, no. 9, pp. 1141
-
1147.

Lagos
-
Quintana, M, Rauhut, R, Lendeckel, W & Tuschl, T 2001, 'Identification of novel genes
coding for small expressed RNAs',
Science
, vol. 294, no. 5543, pp. 853

858.

Lee, RC, Feinbaum,

RL & Ambros, V 1993, 'The C. elegans heterochronic gene lin
-
4 encodes
small RNAs with antisense complementarity to lin
-
14',
Cell
, vol. 75, no. 5, pp. 843

854.

Lewis, BP, Shih, I, Jones
-
Rhoades, MW, Bartel, DP & Burge, CB 2003, 'Prediction of
mammalian mic
roRNA targets',
Cell
, vol. 115, pp.787

798.

Liu, B, Li, J & Tsykin, A 2009a, 'Discovery of functional miRNA

mRNA regulatory modules
with computational methods',
Journal of Biomedical Informatics
, vol. 42, no. 4, pp. 685
-
691.

Liu, B, Li, J, Tsykin, A, Liu,
L, Gaur, A & Goodall, G 2009b, 'Exploring complex miRNA
-
mRNA interactions with Bayesian networks by splitting
-
averaging strategy',
BMC
Bioinformatics
, vol. 10, no. 1, pp. 408.

Lu, J, Getz, G, Miska, EA, Alvarez
-
Saavedra, E, Lamb, J, Peck, D, Sweet
-
Cordero,

A, Ebert,
BL, Mak, RH, Ferrando, AA, Downing, JR, Jacks, T, Horvitz, HR & Golub, TR 2005,
‘MicroRNA expression profiles classify human cancers’,
Nature
, vol. 435, pp. 834

838.

26


Mole, DR, Blancher, C, Copley, RR, Pollard, PJ, Gleadle, JM, Ragoussis, J & Rat
cliffe, PJ 2009,
'Genome
-
wide Association of Hypoxia
-
inducible Factor(HIF)
-
1α and HIF
-
2α DNA Binding with
Expression Profiling of Hypoxia
-
inducible Transcripts',
The Journal of Biological Chemistry
,

The American Society for Biochemistry and Molecular Biolo
gy
, vol. 284, no. 25, pp. 16767
-
16775.

Moore, P. B. & Steitz, T. A. 2002, 'The involvement of RNA in ribosome function',
Nature
, vol.
418, no. 6894, pp. 229
-
235.

National Institutes of Health, “Talking Glossary of Genetic Terms”,
National Human Genome
Research Institute
, Viewed on June 12 2010, <http://www.genome.gov/glossary/?id=123>

Paustian, T & Roberts, G 2006, Through the microscope: a look at all things small,
Lulu.com
.

Roac
h, RC, Wagner, PD & Hackett, PH 2001
,
Hypoxia: from genes to the bedside
,
Kluwer
Academic/Plenum Publishers, New York.

Sebastiani, P, Abad, MM & Ramoni, MF 2004, ' Bayesian networks for genomic analysis'.

Schickel, R, Boyerinas, B, Park, SM &

Peter, ME 2008, ‘MiRNAs: key players in the immune
system, differentiation, tumorigenesis and cell death’, Oncogene, vol. 27, pp. 5959

5974.

Shyu, A
-
B, Wilkinson, MF & van Hoof, A 2008, 'Messenger RNA regulation: to translate or to
degrade', EMBO Journal
, vol. 27, no. 3, pp. 471
-
481.

Tran, DH, Satou, K & Ho, TB 2008, 'Finding microRNA regulatory modules in human genome
using rule induction',
BMC Bioinformatics
, vol. 9, no. 12, pp. S5.

Wang, G, Jiang, B, Rue, E & Semenza, G 1995, 'Hypoxia
-
Inducible Factor

1 is a Basic
-
Helix
-
Loop
-
Helix
-
PAS Heterodimer Regulated by Cellular O2 Tension',
PNAS
, vol. 92, no. 12, June 6,
1995, pp. 5510
.

27


Wienholds, E & Plasterk, RHA 2005, 'MicroRNA function in animal development',
Federation
of European Biochemical Societies Lett
er
, vol. 579, no. 26, pp. 5911
-
5922.

Wightman, B, Ha, I & Ruvkun, G 1993, 'Posttranscriptional regulation of the heterochronic gene
lin
-
14 mediates temporal pattern formation in C. elegans',
Cell
, vol. 75, no. 5, pp. 855

862.

Yoon, S & De Micheli, G 2005,
'Prediction of regulatory modules comprising microRNAs and
target genes',
Bioinformatics
, vol. 21, no. 2, pp. ii93
-
ii100.