to gene co-expression Marie VERBANCK (Agrocampus Ouest / - - PowerPoint PPT Presentation

Marie VERBANCK (Agrocampus Ouest / CNRS-UMR6625, France) Sébastien LÊ (Agrocampus Ouest / CNRS-UMR6625, France)

Integration of biological knowledge related to gene co-expression

Compstat

The data

<Experiment>

Chickens (x27): physiological state

• N: fed (ad libitum access to food) (x6)
• J16: 16-hour fasting (x5)
• J16R5: 16-hour fasting + 5-hour renutrition phase (x7)
• J16R16: 16-hour fasting + 16-hour renutrition phase (x9)
• fatty acid concentrations (hepatic and plasmatic)
• gene expressions (selection)

The data

<Experiment>

Chickens (x27): physiological state

• N: fed (ad libitum access to food) (x6)
• J16: 16-hour fasting (x5)
• J16R5: 16-hour fasting + 5-hour renutrition phase (x7)
• J16R16: 16-hour fasting + 16-hour renutrition phase (x9)
• fatty acid concentrations (hepatic and plasmatic)
• gene expressions (selection)

What are the mechanisms implemented in reply to fasting?

‘-omics’ data

1 j1 J1 1

i I

1 j2 J2

<Merged data tables>

The data, the expectations

< Fatty acid concentrations > < Gene expressions >

‘-omics’ data

1 j1 J1 1

i I

1 j2 J2

<Merged data tables>

The data, the expectations

< Fatty acid concentrations > < Gene expressions > <Expectations>

To provide an help on the functional interpretation in an exploratory multivariate analysis framework

Exploratory multivariate analysis framework

The multitude of gene expressions is projected onto the the correlation circle uninterpretable

Exploratory multivariate analysis framework

The multitude of gene expressions is projected onto the the correlation circle uninterpretable

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Assembly of genes into modules Interpretation at the level of the groups

Exploratory multivariate analysis framework

Modules

<MODULES of GENES>

‘-omics’ data

1 j1 J1 1 j2 J2 1

i I

Modular approach  Simultaneous interpretation of several genes through the supplementary groups projected onto the group representation

‘-omics’ data

1 j1 J1 1 j2 J2

M1 <MODULES of GENES>

1

i I

Modules

Modular approach  Simultaneous interpretation of several genes through the supplementary groups projected onto the group representation

‘-omics’ data

1 j1 J1 1 j2 J2

M1 M2 <MODULES of GENES>

1

i I

Modules

Modular approach  Simultaneous interpretation of several genes through the supplementary groups projected onto the group representation

‘-omics’ data

1 j1 J1 1

i I

1 j2 J2

M1 M2 M3 …..

Modules

<MODULES of GENES>

Modules

Modular approach  Simultaneous interpretation of several genes through the supplementary groups projected onto the group representation

1 j1 J1 1

i I

1 j2 J2

M1 M2 M3 ….. <MODULES of GENES>

Modules

<z1,z1> <z2,z2>

I2

R

Modular approach

Where z1 denotes the first main axis of variability among the individuals

1 j1 J1 1

i I

1 j2 J2

M1 M2 M3 ….. <MODULES of GENES>

Modules

M1

<z1,z1> <z2,z2>

M1 M1

Scalar products matrices between chickens

I2

R

Modular approach

1 j1 J1 1

i I

1 j2 J2

M1 M2 M3 ….. <MODULES of GENES>

Modules

M1

<z1,z1> <z2,z2>

M1 M1

Scalar products matrices between chickens

I2

R

Modular approach

1 j1 J1 1

i I

1 j2 J2

M1 M2 M3 ….. <MODULES of GENES>

Modules

M1

<z1,z1> <z2,z2>

M1 M1

Scalar products matrices between chickens

I2

R Lg(z1,M1)

Modular approach

1 j1 J1 1

i I

1 j2 J2

M1 M2 M3 ….. <MODULES of GENES>

Modules

M1

<z1,z1> <z2,z2>

M1 M1

Scalar products matrices between chickens

I2

R Lg(z1,M1)

Modular approach

j j g

K z K z L

• f

comp. princ. 1 the is 1 ) , (

st

 

Biological knowledge

< “a priori” information >

Description of genes and genes products

• Cellular Component
• Molecular Function
• Biological Process (BP)

Gene Ontology

Genes could be grouped by GO BP terms

Our Approach

< “a posteriori” information >

1 . . . . j . . . . q 1 . . . . j' . . . . p 1 : :

Genes

i

gij mij'

: : n

Terms Expression profiles(microarrays)

G M

G: Contingency table gij = 1 if the gene i belongs to the process j 0 if not M: Quantitative data frame Transpose of the table microarrays x genes, the data being centered by row

< “a posteriori” information >

Our Approach

Our Approach

Two genes are close in this space if: 1- They are involved in the same biological processes 2- They are co-expressed 3- They are situated at a similar level of the regulatory network

Construction of a space with a new distance between the genes: < “a posteriori” information >

Our Approach

Two genes are close in this space if: 1- They are involved in the same biological processes 2- They are co-expressed 3- They are situated at a similar level of the regulatory network

Construction of a space with a new distance between the genes:

 The two genes must be associated to the same terms  Matrix of the terms

< “a posteriori” information >

Our Approach

Two genes are close in this space if: 1- They are involved in the same biological processes 2- They are co-expressed 3- They are situated at a similar level of the regulatory network

Construction of a space with a new distance between the genes:

 The two genes must be associated to the same terms  Matrix of the terms  The two gene expressions must induce the same structure on the individuals  Gene expressions data frame

< “a posteriori” information >

Our Approach

Two genes are close in this space if: 1- They are involved in the same biological processes 2- They are co-expressed 3- They are situated at a similar level of the regulatory network  The number of processes the gene is involved in could determine its level in the network  Weighting

Construction of a space with a new distance between the genes:

 The two genes must be associated to the same terms  Matrix of the terms  The two gene expressions must induce the same structure on the individuals  Gene expressions data frame

< “a posteriori” information >

Our Approach

Two genes are close in this space if: 1- They are involved in the same biological processes 2- They are co-expressed 3- They are situated at a similar level of the regulatory network  The number of processes the gene is involved in could determine its level in the network  Weighting

Construction of a space with a new distance between the genes:

Canonical Correspondence Analysis  The two genes must be associated to the same terms  Matrix of the terms  The two gene expressions must induce the same structure on the individuals  Gene expressions data frame

< “a posteriori” information >

Our Approach

• 2
• 1

1 2

• 1

1 2 3

RIGG06437 RIGG16397 RIGG02005 RIGG15080 RIGG02523 RIGG01056 RIGG20074 RIGG14063 RIGG03937 RIGG15083 RIGG08970 RIGG20299 RIGG06749 RIGG11656 RIGG05667 RIGG06682 RIGG17220 RIGG02730 RIGG07959 RIGG09550 RIGG00015 RIGG18148 RIGG03089 RIGG10681 RIGG16849 RIGG13646 RIGG15481 RIGG00865 RIGG18276 RIGG12903 RIGG19372 RIGG19793 RIGG15064 RIGG05911 RIGG14333 RIGG11544 RIGG02231 RIGG18271 RIGG04625 RIGG08140 RIGG08865

Representation

• f

the genes

• nto

the canonical variables

Our Approach

• 2
• 1

1 2

• 1

1 2 3

RIGG06437 RIGG16397 RIGG02005 RIGG15080 RIGG02523 RIGG01056 RIGG20074 RIGG14063 RIGG03937 RIGG15083 RIGG08970 RIGG20299 RIGG06749 RIGG11656 RIGG05667 RIGG06682 RIGG17220 RIGG02730 RIGG07959 RIGG09550 RIGG00015 RIGG18148 RIGG03089 RIGG10681 RIGG16849 RIGG13646 RIGG15481 RIGG00865 RIGG18276 RIGG12903 RIGG19372 RIGG19793 RIGG15064 RIGG05911 RIGG14333 RIGG11544 RIGG02231 RIGG18271 RIGG04625 RIGG08140 RIGG08865

These genes are brought together on the plan coming from CCA: they induce the same structure

• nto the expression

profiles (correlation of 0.94)

Representation

• f

the genes

• nto

the canonical variables

Our Approach

• 2
• 1

1 2

• 1

1 2 3

RIGG06437 RIGG16397 RIGG02005 RIGG15080 RIGG02523 RIGG01056 RIGG20074 RIGG14063 RIGG03937 RIGG15083 RIGG08970 RIGG20299 RIGG06749 RIGG11656 RIGG05667 RIGG06682 RIGG17220 RIGG02730 RIGG07959 RIGG09550 RIGG00015 RIGG18148 RIGG03089 RIGG10681 RIGG16849 RIGG13646 RIGG15481 RIGG00865 RIGG18276 RIGG12903 RIGG19372 RIGG19793 RIGG15064 RIGG05911 RIGG14333 RIGG11544 RIGG02231 RIGG18271 RIGG04625 RIGG08140 RIGG08865

These genes are brought together on the plan coming from CCA: they induce the same structure

• nto the expression

profiles (correlation of 0.94) Those two genes are moved apart by the CCA: they induce different structures onto the expression profiles (correlation of 0.11)

Representation

• f

the genes

• nto

the canonical variables

Our Approach

Objective: to constitute groups of genes. Classification of the genes according to their coordinates on the canonical variables (150 groups).

Our Approach

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Objective: to constitute groups of genes. Classification of the genes according to their coordinates on the canonical variables (150 groups). Projection onto the groups’ representation

Our Approach

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Objective: to constitute groups of genes. Classification of the genes according to their coordinates on the canonical variables (150 groups). Projection onto the groups’ representation

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Interpretation of a group Neighboring terms Terms which the genes of the group are associated to Gene functions

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Interpretation

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Neighboring terms

• regulation of catabolic process
• regulation of actin cytoskeleton
• rganization
• histone modification
• regulation of translation
• regulation of cell growth
• covalent chromatin modification

Interpretation

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Neighboring terms

• regulation of catabolic process
• regulation of actin cytoskeleton
• rganization
• histone modification
• regulation of translation
• regulation of cell growth
• covalent chromatin modification

Interpretation

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Neighboring terms

• regulation of catabolic process
• regulation of actin cytoskeleton
• rganization
• histone modification
• regulation of translation
• regulation of cell growth
• covalent chromatin modification

Associated terms

• proteolysis
• transcription
• glycerol metabolic process
• translation
• lipid metabolic process
• fatty acid metabolic process
• gene expression
• glycerol catabolic process

Interpretation

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Neighboring terms

• regulation of catabolic process
• regulation of actin cytoskeleton
• rganization
• histone modification
• regulation of translation
• regulation of cell growth
• covalent chromatin modification

Gene functions

• Calpain-1-cata-protease: Proteolysis
• Fructose 1,6 biphosphatase: Neoglucogenesis
• UDP glucusyltransferease: Lactose synthesis
• 3 ketoacyl-coA: Fatty acid oxidation

Associated terms

• proteolysis
• transcription
• glycerol metabolic process
• translation
• lipid metabolic process
• fatty acid metabolic process
• gene expression
• glycerol catabolic process

Interpretation