Multifactorial Disease Genomics - week 8

Question 1

what are single gene disorders called?

Answer 1

mongenic and mendelian

Question 2

features of single gene disorders?

Answer 2

rare, specific pattern of inheritance in family, disease caused by pathogenic variant

Question 3

example of single gene disorder

Answer 3

Cystic Fibrosis – mutation in a single gene (recessive or dominant) that will always have the same phenotypic affect

Question 4

The majority of morbidity and mortality in 21st Century Britain is the result of

Answer 4

genetic and environmental factors

Question 5

Multifactorial are a Combination of both

Answer 5

genes and non-genetic(environmental) factors that determine disease risk
o Diseases such as Non – communicable diseases (type 2 diabetes, heart disease and cancers)

Question 6

How do we know common traits have a genomic component? (twin studies + family studies)

Answer 6

GOLD STANDARD: TWIN STUDIES
Do both twins show the same characteristic or trait? Comparing MZ/DZ twins can give evidence for genetic and/or environmental influences

Question 7

MZ twins share

Answer 7

all their genes and environment (genetically identical)

o Much more physically similar than DZ twins

Question 8

DZ twins share

Answer 8

50% genes and environment

Question 9

By looking at various traits i.e. body fat distribution, height and weight and by comparing DZ and MZ twins you can see how

Answer 9

concordant the two are.

Question 10

Concordance = For a trait like height monozygotic twins are

Answer 10

95% concordant compared to 52% of dizygotic twins so height is a highly heritable trait – estimated that it’s about 80% heritable

Question 11

Concordance = BMI is slightly

Answer 11

less heritable but still heritable as concordance higher in monozygotic than in dizygotic

Question 12

Parent offspring correlations

Answer 12

Another way of measuring heritability is looking at correlation between parents and offspring for a particular trait

Question 13

How do you conduct parent offspring correlations?

Answer 13

Plot average of parents vs offspring’s phenotypes

If no genetic affect of this trait then you’d expect both phenotypes to be random

If there is a correlation then if gives you a slope of the line – indication that the trait is heritable

Look at the correlation using the “slope” of the line

Question 14

The Genetic basis of common polygenic disorders

Answer 14

(What is the difference between single gene disorders and common disease with partial genetic component)

Question 15

How do we measure genetic basis of common polygenic disorders? background

Answer 15

These common polygenic diseases are likely to have many DNA changes, each predisposing to disease (type 2 diabetes, cardiovascular disease, schizophrenia, …) (unlike monogenic diseases – you can say that a particular variant is causative)

Question 16

How do we measure genetic basis of common polygenic disorders?

Answer 16

So you could look at a case control study – looking at a particular variant in cases, present in 33% of cases and only 16% of the controls – indication that this particular variants is associated with your disease.

Question 17

For complex disease traits a gene variant is present more

Answer 17

frequently in cases than controls

Question 18

Variant status (genotype) provides

Answer 18

probability of disease status

Question 19

How do we measure genetic basis of common polygenic disorders? another way of measuring

Answer 19

Common polygenic quantitative traits many DNA changes, each influencing levels of a trait (height, BMI, blood pressure, CRP, LDL, HDL, triglycerides, fasting glucose,… - no case control).

For quantitative traits, compare the mean trait value (mean BMI) between the three genotype groups. If the 3 groups are different then that’s an indication also that there’s an association with a particular locus.

Question 20

Before 2007

Answer 20

Candidate Gene studies: Before 2007 you could do association studies but gene by gene in a candidate gene method
o i.e. gene on chromosome 12, gene z involved in height I’m going to test it for an association – if no association then have to test the other 24,999 genes

Question 21

2007: Genome-wide association studies

Answer 21

Genotype and test thousands of SNPs across genome (hypothesis free approach – no prior knowledge of the trait needed)

Question 22

“Genome-wide significance” example BMI

Answer 22

Here’s the quantitative trait for BMI, 3 genotype groups, and can test whether the difference in mean BMI is significant different in the 3 BMI groups using linear regression.
Association statistics derived from regression analyses

Question 23

“Genome-wide significance” example BMI Linear regression looking at

Answer 23

deviation from a linear relationship

Question 24

“Genome-wide significance” example BMI: vertical black line

Answer 24

average change in BMI per BMI increasing allele

Question 25

“Genome-wide significance” example BMI: Generate a P-value returned for each SNP tested:

Answer 25

probability of observing an association by chance when no association actually exists

Question 26

“Genome-wide significance” example BMI: common significance threshold?

Answer 26

Common to use P=0.05 as significance threshold to be classified as “significant”
5% probability (1 in 20) of seeing the observation by chance

Question 27

“Genome-wide significance” example BMI: p = 0.05

Answer 27

There are around 1M independent common SNPs in genome after accounting for linkage disequilibrium
P=0.05 would mean we would expect 50,000 associations at that level of significance by chance!!!
Solution: correct for number of test: 0.05/1,000,000 = 5x10-8 (known as Bonferroni correction)
(P value divided by number of tests doing (estimate 1 million))

Question 28

GWAS results Manhattan plots show

Answer 28

Manhattan Plot showing association of genetic variants across the genome with type 2 diabetes

Question 29

Each point represents a

Answer 29

SNP.

Question 30

The position of each SNP relates to:

Answer 30

1) its position in the genome (X-axis)

2) It’s strength of association against T2D when tested using x amount of T2D patients

Question 31

GWAS results show common variation in

Answer 31

at least 97 gene regions predispose to obesity

Question 32

There is a correlation between

Answer 32

the size of the effect of genetic variation and frequency in population

Question 33

Mendelian category

Answer 33

There are many diseases that fall in the mendelian monogenic disorders which are very rare but have a high penetrance so most of the time will cause exactly the same disease in everyone who carries it

Question 34

GWAS variants

Answer 34

GWAS variants fall in the category on the right – these are common variants at a frequency of greater than 0.1% and these generally have a much smaller effect so need multiple variants

Question 35

Middle category

Answer 35

Now starting to understand more about the genes in the category in the middle – relatively uncommon (low frequency) but have an intermediate penetrance so fairly big affect but not near 100% penetrance

Question 36

The other categories – less known

Answer 36

o Low penetrance and very rare = very hard to identify

o High penetrance and very common = uncommon that these would exist as generally they would be selected against.

Question 37

Limitations of GWAS

Answer 37

larger sample sizes required,
variants have small effect sizes,
causal pathway, gene and variant not always determined,
method based on SNP variation
ethnic groups studied to date have been limited
Pleiotropy is common

Question 38

Limitations of GWAS: larger sample sizes required

Answer 38

no plateau effect seen yet (the larger the sample size the better – as you increase sample size you see more and more variants, so lots of traits have huge numbers of genetic variants all contributing a small amount to the phenotype, the more you find the lower the effect size)

Question 39

Limitations of GWAS: variants have small effect sizes

Answer 39

Each individual variant has a very small effect sizes so might only affect height by only a few centimetres

Question 40

Limitations of GWAS: causal pathway, gene and variant not always determined

Answer 40

Causal pathway, gene and variant not always determined – to be able to identify the causal relationship then functional work required

Question 41

Limitations of GWAS: method based on SNP variation

Answer 41

The method is based on SNP variation and we know that SNPs do not cover all variation in the genome, e.g. There are larger copy number variations and multi-allelic variants which won’t be covered well by SNPS

Question 42

Limitations of GWAS ethnic groups studied to date have been limited

Answer 42

Ethnic groups studied to date have been limited – do results apply across all groups?
• GWAS data existing is very biased towards populations of European ancestry
• This is an active area of research so difficult to apply findings to other ethnic groups

Question 43

Limitations of GWAS: Pleiotropy

Answer 43

• Pleiotropy is common – variants associated with multiple traits
• (sometimes this can help but sometimes makes it more difficult to determine the effect of a particular variant is)

Question 44

GWAS benefits:

Answer 44

new technology
determine causal relationships between traits and disease
predicting who will get disease
drug discovery

Question 45

GWAS benefits: new technology

Answer 45

• Huge benefits including understanding new biology and this is the case for all of the GWAS studies as it’s uncovered new biology as the approach doesn’t look at candidate genes it’s hypothesis free so you can use GWAS to find new areas of biology associated with your trait

• It allows you to dissect the biology of type 2 diabetes
o Type 2 diabetes – a number of genes near common variants were identified and by looking at what the genes are, they’re expression in different tissues you can start to partition them into different groups
 Transcription factors, appetite control, BMI, beta cell effect
 So it helps to understand the fundamental etiology of type 2 diabetes

Question 46

GWAS benefits: determine causal relationships between traits and disease

Answer 46

Second important area is being able to identify causal relationship between traits and disease

Question 47

GWAS Problem: correlation does not equal causation

Answer 47

• just because 2 things are associated doesn’t mean it’s causative i.e.
• people in Florida are more likely to have Alzheimer’s. Does that mean that living in Florida causes you to develop Alzheimer’s?
• Probably not – more likely that older people move to Florida and older people develop Alzheimer’s – this is a confounding factor

• Association between coffee drinking and pancreatic cancer, people who smoke are also more likely to drink more coffee and smoking is a risk factor for pancreatic cancer. These observational studies are often very difficult to unpick exactly what the causal relationship is.

Question 48

Mendelian randomisation can replace a randomised control trial

Answer 48

• Example of using genetics to solve this problem = technique called Mendelian randomisation which can be used to replace RCT (gold standard)
• Association between high BMI and likelihood of Type 2 diabetes, but you want to understand if this is a causal relationship. Could it be confounded by lack of exercise, which we known is associated with both BMI and Type 2 diabetes?
• So to understand if BMI is directly causing an increased risk of type 2 diabetes then you could have a RCT where you have a bunch of people with high BMI and then a bunch of people with low BMI, keeping all other conditions exactly the same, to see who develops type 2 diabetes = very difficult to do!!
• So instead can identify genetic variants associated with BMI (around 100 genetic variants associated with BMI) which are not associated with exercise (confounder) so you can check if the genetic variants are directly associated with type 2 diabetes and if they are then it suggests that BMI is on the causal pathway to developing type 2 diabetes.

Question 49

GWAS benefits: predicting who will get disease

Answer 49

• Heart disease has been one of the leaders of this (type 1 diabetes shown too)
• Can combine all the genetic variants each having a small affect into a polygenic risk score

Question 50

Combining genetic variant into polygenic risk scores (PRS) looks particularly promising for cardiovascular disease

Answer 50

• Middle panel = Polygenic risk scores for a series of cases with coronary artery disease compared to control, the risk score is significantly higher for cases compared to controls
• Panel on right = shows individuals with different genetic risk scores, as you go along the X axis they have an increasing genetic risk for cardiovascular disease and the Y axis shows prevalence of the disease (proportion of the individuals who develop it)
o so as the genetic risk increases, the likelihood of them developing the condition also increases.
• People are born with their genetic component for heart disease so we can determine what individual risk is for heart disease
• Panel on left = As you plot it, it shows that if you’re in the top 5% of the population for genetic risk you have a fivefold increased risk of developing coronary artery disease
o This could be an opportunity for targeted intervention in people with the risks or adding to conventional risk factors

Question 51

GWAS benefits: drug discovery

Answer 51

• Number of studies going backwards have shown that by looking at well-known drugs for certain disorders you can identify targets of the drugs by looking at the GWAS hits which have come up for those particular traits
• Drug companies can save a lot of money in drug development by using genetics to allow them to target particular regions.