Transcript Gene expression in CHILD and biological and technical variation

Aim of this session
• Introduce finding in gene expression
• Discuss biology and technology behind it
Gene expression in
CHILD
1. The data
2. Introduction to the phenomenon
3. Analysis of the phenomenon
Ethnicity Mother
0%
20%
Caucasian
40%
South-east Asian
60%
First nation
80%
South Asian
Black
100%
Other
BMI Mother
BMI
The neonate
Birth
weight
Gestational age
500 neonates
35000
probes
1. The data
2. Introduction to the phenomenon
3. Analysis of the phenomenon
Ingenuity Pathway Analysis (IPA)
Ingenuity Pathway Analysis (IPA)
• Finds patterns in a dataset, using 1000s of
pre-defined gene lists.
Ingenuity
Pathway
Analysis
Gene 1
Gene 2
…
…
…
Gene 576
Gene 577
Ingenuity
Pathway
Analysis
Gene 1
Gene 2
…
…
…
Gene 576
Gene 577
Ingenuity
Pathway
Analysis
Gene 1
Gene 2
…
…
…
Gene 576
Gene 577
A weak, but super
consistent differential
expression between
participants (p-value=0)
Gene expression patterns
resemble those
of a virus-infected cell
“Virus Z-Score”
Ingenuity
Pathway
Analysis
Gene 1
Gene 2
…
…
…
Gene 576
Gene 577
Sort by consistency,
top 50% as Z-score
Correlation with virus Z-score
1. The data
2. Introduction to the phenomenon
3. Analysis of the phenomenon
Theory
Conditions were not sterile.
CHILD
“Virus
Z-score”
Study centres
Replication in START,
and online dataset
Theory
Conditions were not sterile.
Theory
Plates were analysed in differential conditions.
Conditions were not sterile.
Expression
(log2)
FURIN expression
Expression
(log2)
Kruskal-Wallis: p=3e-14
ZC3HAV1 expression
Expression
(log2)
Kruskal-Wallis: p=3e-12
Theory
Plates were analysed in differential conditions.
Conditions were not sterile.
Theory
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Theory
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Time gap between birth and blood sampling caused
transcriptomic alterations.
Minutes
Minutes
Birth
<10 min
Placenta
<10 min
Cord blood
CHILD
“Virus
Z-score”
Minutes from birth to blood collection (log10)
Theory
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Time gap between birth and blood sampling caused
transcriptomic alterations.
“Virus Z-score” is not
associated with
•
•
•
•
•
•
•
•
•
•
•
Ethnicity
Gestational age or birth weight
Gender of the child
Nutrition or smoking
Participant number
Day/night, or winter/spring/summer/autumn
Mode of delivery
City of delivery
Appearance of placenta
Time between birth and blood collection
Global methylation
Theory
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Time gap between birth and blood sampling caused
transcriptomic alterations.
Endogenous retroviruses (ERVs) migrated into the blood
vessel, influencing gene expression.
Theory
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Time gap between birth and blood sampling caused
transcriptomic alterations.
Endogenous retroviruses (ERVs) migrated into the blood
vessel, influencing gene expression.
Theory
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Trophoblasts were included in the needle.
Time gap between birth and blood sampling caused
transcriptomic alterations.
Endogenous retroviruses (ERVs) migrated into the blood
vessel, influencing gene expression.
Findings in top-500 of transcriptome
Gene name
Common name
Esp. found in tissue
GCM1
Glial cells missing homolog 1
Trophoblasts
MBNL3
Muscleblind-like 3
Trophoblasts
MUC6
Mucin 6, oligomeric mucus/gel-forming
Trophoblasts
TMEM158
Transmembrane protein 158
Endometrium SM
ADRA2C
Adrenergic, alpha-2C-, receptor
Endometrium SM / decidua
Criterion: gene has high tissue specificity and sensitivity
Findings in rest of transcriptome
Gene name
Common name
Esp. found in tissue
SLC2A1
GLUT1 (glucose transporter)
Trophoblasts
SLC43A2
Large neutral amino acid transporter
Trophoblasts
SERPINE2
Serpin peptidase inhibitor, clade E
Trophoblasts
GAA
Glucosidase, alpha; acid
Trophoblasts
HEXB
Hexosaminidase B
Trophoblasts
CLIC3
Chloride intracellular channel 3
Trophoblasts, decidua
FOXO4
Forkhead box O4
Trophoblasts
Criteria: gene has high placenta specificity and
sensitivity, gene correlates with others in this table
Theory
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Trophoblasts were included in the needle.
Time gap between birth and blood sampling caused
transcriptomic alterations.
Endogenous retroviruses (ERVs) migrated into the blood
vessel, influencing gene expression.
“Trophoblast Z-Score”
CHILD
“Virus
Z-score”
“Trophoblast
Z-score”
CHILD
“Virus
Z-score”
MUC6 expression (log2)
START
“Virus
Z-score”
MUC6 expression (log2)
Theory
Concentrated results due to varying path of differentiation.
Virus infection pattern merely based on small differences
in baseline antiviral defence.
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Trophoblasts were included in the needle.
Time gap between birth and blood sampling caused
transcriptomic alterations.
Endogenous retroviruses (ERVs) migrated into the blood
vessel, influencing gene expression.
Correlation with virus Z-score
Cell type 1
Cell type 2
Correlation with virus Z-score
Correlation with virus Z-score
Genes of virus Z-score itself
Correlation with virus Z-score
HOX genes,
Keratins,
LGALS12, ELAVL3, MIR1976
Genes of virus Z-score itself
Correlation with virus Z-score
HOX genes, Keratins, LGALS12, ELAVL3, MIR1976
HOX genes, Keratins, LGALS12, ELAVL3, MIR1976
Maron et al.
HOX genes, Keratins, LGALS12, ELAVL3, MIR1976
Maron et al.
Matigian et al.
HOX genes, Keratins, LGALS12, ELAVL3, MIR1976
Maron et al.
Matigian et al.
Merkerova et al.
HOX genes, Keratins, LGALS12, ELAVL3, MIR1976
Maron et al.
Matigian et al.
Merkerova et al.
Yoshida et al.
Yoshida et al.
Theory
Concentrated results due to varying path of differentiation.
Virus infection pattern merely based on small differences
in baseline antiviral defence.
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Trophoblasts were included in the needle.
Time gap between birth and blood sampling caused
transcriptomic alterations.
Endogenous retroviruses (ERVs) migrated into the blood
vessel, influencing gene expression.
Theory
Concentrated results due to varying path of differentiation.
Virus infection pattern merely based on small differences in
baseline antiviral defence.
Normalisation confused biological variation.
Plates were analysed in differential conditions.
Conditions were not sterile.
Trophoblasts were included in the needle.
Time gap between birth and blood sampling caused
transcriptomic alterations.
Endogenous retroviruses (ERVs) migrated into the blood
vessel, influencing gene expression.
Gene expression in cord blood mononuclear cells (CBMCs)
is not biologically relevant.
Correlation with virus Z-score
CHILD
“Virus
Z-score”
FOXO3 expression (log2)
(=key TF, typifying “distinct patterns of gene expression”)
Safety first? Name it:
“Distinct patterns”!
Australia, Nov 2014
Safety first? Name it:
“Distinct patterns”!
Australia, Nov 2014
Dataset is publicly available… and guess what?
WWW
“Virus
Z-score”
“Trophoblast
Z-score”
FOXO4
expression
(log2)
CHILD
START
WWW
HEXB
expression
(log2)
CHILD
START
WWW
Patterns related (+/-) to Virus Z-score
+
RB……
UBQLN…
PSM……
ATP……
…………
ADAM8
SEC16A
PHIP
CYTH1
Pattern
HOXB6
Virus Z-score: HOX……
immune-like
HOXC10
phenotype and
trophoblasts KRT3
KRT……
EPPK1
Cell migration, LGALS12
inhibition of
insulin-signalling ELAVL3
MIR1976
Pattern
Differentiation
into specific
cell types
other than
immune cells
(e.g. neuron,
fibroblast,
adipocyte,
muscle)
Anti-proliferative